What do you think?
Rate this book
Co-Intelligence: Living and Working with AI
AN INSTANT NEW YORK TIMES BESTSELLER
From Wharton professor and author of the popular One Useful Thing Substack newsletter Ethan Mollick comes the definitive playbook for working, learning, and living in the new age of AI
Something new entered our world in November 2022 — the first general purpose AI that could pass for a human and do the kinds of creative, innovative work that only humans could do previously. Wharton professor Ethan Mollick immediately understood what ChatGPT after millions of years on our own, humans had developed a kind of co-intelligence that could augment, or even replace, human thinking. Through his writing, speaking, and teaching, Mollick has become one of the most prominent and provocative explainers of AI, focusing on the practical aspects of how these new tools for thought can transform our world.
In Co-Intelligence, Mollick urges us to engage with AI as co-worker, co-teacher, and coach. He assesses its profound impact on business and education, using dozens of real-time examples of AI in action. Co-Intelligence shows what it means to think and work together with smart machines, and why it's imperative that we master that skill.
Mollick challenges us to utilize AI's enormous power without losing our identity, to learn from it without being misled, and to harness its gifts to create a better human future. Wide ranging, hugely thought-provoking, optimistic, and lucid, Co-Intelligence reveals the promise and power of this new era.
From Wharton professor and author of the popular One Useful Thing Substack newsletter Ethan Mollick comes the definitive playbook for working, learning, and living in the new age of AI
Something new entered our world in November 2022 — the first general purpose AI that could pass for a human and do the kinds of creative, innovative work that only humans could do previously. Wharton professor Ethan Mollick immediately understood what ChatGPT after millions of years on our own, humans had developed a kind of co-intelligence that could augment, or even replace, human thinking. Through his writing, speaking, and teaching, Mollick has become one of the most prominent and provocative explainers of AI, focusing on the practical aspects of how these new tools for thought can transform our world.
In Co-Intelligence, Mollick urges us to engage with AI as co-worker, co-teacher, and coach. He assesses its profound impact on business and education, using dozens of real-time examples of AI in action. Co-Intelligence shows what it means to think and work together with smart machines, and why it's imperative that we master that skill.
Mollick challenges us to utilize AI's enormous power without losing our identity, to learn from it without being misled, and to harness its gifts to create a better human future. Wide ranging, hugely thought-provoking, optimistic, and lucid, Co-Intelligence reveals the promise and power of this new era.
Audible Audio
First published April 2, 2024
4223 people are currently reading
23750 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 1,687 reviews
May 28, 2024
This is an interesting and useful book, and if you're an academic or educator who's trying to find a good strategy for dealing with the AI revolution then I strongly recommend reading it. Everyone I know in this area is having problems: the whole landscape has changed overnight now that students have access to ChatGPT, Gemini, Claude and similar platforms. People have tried banning students from using them, which doesn't work, and they've tried ignoring the problem and hoping things will somehow just sort themselves out, which works even less. Ethan Mollick, a US professor of business studies, realised at an early stage that a more coherent approach was to embrace the new technology and learn to use it as well as possible. Even if you think this is absolutely the wrong thing to do, you should read about his experiences to see what embracing it might involve; a great many of your colleagues will be taking this route.
Mollick has some extremely sensible advice to offer; at times it may seem paradoxical, but we're in a paradoxical situation. He repeatedly says that AIs are not human beings - in particular, he considers it clear that they are not conscious - but he also says that you will be better at using them if you treat them as though they were human beings. We don't yet have a good way to conceptualise AIs, and they behave much more like people than like the machines we're used to. He says in particular that it will be helpful to tell the AI what kind of person you want them to resemble, and gives nice examples of how useful it can be to have an AI who's been instructed to be, inter alia, the object of a sales pitch, a critical reader of your essay, or a potentially interested but uninformed member of the public. He repeatedly urges you to be sceptical and gather information. The abilities of current-generation AIs are often hard to imagine until you've actually tried them on a given task: a striking example which many people have discovered independently is that ChatGPT can't spot a one move threat at Tic-Tac-Toe, but if asked will immediately write you a Python script that plays the game perfectly. Who would have predicted that?
Mollick not merely encourages but requires his students to use AI, and shows you how much they can get out of it. The teacher can dramatically raise the bar, since every member of the class now has a competent virtual team working for them. You don't have to limit business students to describing the website they would build; even if they can't code, you ask them actually to build a prototype, relying on the AI to take care of the implementation. Mollick says he can't code, but found this worked for him, and he's passing on his experience. The technology is improving all the time, and things which were difficult or impossible a few months ago may be quite feasible in the next release. ChatGPT 3.5, which took the world by storm in November 2022, is already starting to feel quaint and old-fashioned. When the voice-enabled version of ChatGPT-4o becomes generally available (looks like it will be Q3/Q4 2024), we'll hit another inflection point. Video capabilities are evidently in the pipeline, as are other less in-your-face but possibly even more important improvements related to planning and reasoning. The impact on education will be revolutionary. Mollick quotes well-known studies showing that personal tuition gives students a two-sigma advantage, moving an average student to the 98th percentile. We are within sight of a world where all students in all but the poorest countries will be able to get expert personal tuition in all subjects.
Where is this going long-term? Mollick honestly says he doesn't know, and there's little point in speculating; what makes sense is to keep up with events and understand what new options are materialising, whether we like them or not. It's possible that AI development will plateau or at least start levelling off. But that isn't what it looks like right now. It's not accidental that we see the Apple of Eden on the front cover.
Such extraordinary times we're living in.
Mollick has some extremely sensible advice to offer; at times it may seem paradoxical, but we're in a paradoxical situation. He repeatedly says that AIs are not human beings - in particular, he considers it clear that they are not conscious - but he also says that you will be better at using them if you treat them as though they were human beings. We don't yet have a good way to conceptualise AIs, and they behave much more like people than like the machines we're used to. He says in particular that it will be helpful to tell the AI what kind of person you want them to resemble, and gives nice examples of how useful it can be to have an AI who's been instructed to be, inter alia, the object of a sales pitch, a critical reader of your essay, or a potentially interested but uninformed member of the public. He repeatedly urges you to be sceptical and gather information. The abilities of current-generation AIs are often hard to imagine until you've actually tried them on a given task: a striking example which many people have discovered independently is that ChatGPT can't spot a one move threat at Tic-Tac-Toe, but if asked will immediately write you a Python script that plays the game perfectly. Who would have predicted that?
Mollick not merely encourages but requires his students to use AI, and shows you how much they can get out of it. The teacher can dramatically raise the bar, since every member of the class now has a competent virtual team working for them. You don't have to limit business students to describing the website they would build; even if they can't code, you ask them actually to build a prototype, relying on the AI to take care of the implementation. Mollick says he can't code, but found this worked for him, and he's passing on his experience. The technology is improving all the time, and things which were difficult or impossible a few months ago may be quite feasible in the next release. ChatGPT 3.5, which took the world by storm in November 2022, is already starting to feel quaint and old-fashioned. When the voice-enabled version of ChatGPT-4o becomes generally available (looks like it will be Q3/Q4 2024), we'll hit another inflection point. Video capabilities are evidently in the pipeline, as are other less in-your-face but possibly even more important improvements related to planning and reasoning. The impact on education will be revolutionary. Mollick quotes well-known studies showing that personal tuition gives students a two-sigma advantage, moving an average student to the 98th percentile. We are within sight of a world where all students in all but the poorest countries will be able to get expert personal tuition in all subjects.
Where is this going long-term? Mollick honestly says he doesn't know, and there's little point in speculating; what makes sense is to keep up with events and understand what new options are materialising, whether we like them or not. It's possible that AI development will plateau or at least start levelling off. But that isn't what it looks like right now. It's not accidental that we see the Apple of Eden on the front cover.
Such extraordinary times we're living in.
January 9, 2026
I wanted to love this book -- and while it's funny and humorous, it's basically a long blog post or online message board discussion. The conversations with AI are amusing - but not substantive.
The author talks about anthropomorphization -- and dismisses people talking to their pets as the same as talking to their cars or other inanimate objects. And yet - he wants to make a case for AI being some kind of "alien mind" - when he's making a stronger case for it as a mirror of our own expectations.
Throughout the book he talks about how AI is filling in the most probable or likely response -- so if you say "To be or not ___ " it's going to say "to be" since that is the most probable. If you say "The alien ate a banana because ___" you'll get more variety every time you ask the question because it's not a topic or subject reflected in the AI training corpus.
AI is just another way for us to summarize back all the stuff we've written/created with words or images -- to us. It's more like a fun-house mirror than an intelligence.
My cats are more intelligent and capable of communication than an AI -- so where does this leave us?
Humans want to imagine that AI is more intelligent because it sounds like us and tells us what we want to hear. AI is very conformist - it will rewrite and reshape your thoughts and ideas to make them more appealing based upon the corpus of information it has already digested. It will align your communication to meet the greatest common denominator. You can ask it questions and it will tell you to analyze "Heart of Darkness" vis a vis Jane Austen's marriage plots -- and come up with some reasonably interesting points based on what it expects are most probable points from what we have already fed it. NOT because it's thinking of these things. Rather, AI acts as a sophisticated tool for summarizing and regurgitating information in a manner that aligns with human expectations.
I'd like to see a deeper exploration of AI's capabilities and limitations beyond surface-level interactions examining the fundamental nature of AI and its true intelligence compared to human and even animal cognition.
The author quotes someone who says that humans are just a historical point in the evolution of "intelligence" as if that is an autonomous being or subject. I'd contend that it is not -- we aren't merely vessels for "intelligence." There's no hierarchy of value based on "intelligence."
However, I do appreciate the author's optimism that AI will remove the boring/redundant tasks from human labor and his call to consider universal basic income and healthcare as real needs as humans move toward an era where we should focus our attention on tasks and activities that are more intrinsically rewarding to us and which do not require a 40 hour (or more) work week.
The mythical 4 day work week has been promised for decades, the practicalities of achieving a reducing "full time" work weeks remain uncertain. Striking a balance between dystopian and utopian visions of AI's impact on society is crucial for informed decision-making.
SOURCE URL: http://www.livegreenwearblack.com/202...
© Jennifer R Clark. This work is licensed under a Creative Commons Attribution 4.0 International License. You may share and adapt this content with proper attribution.
The author talks about anthropomorphization -- and dismisses people talking to their pets as the same as talking to their cars or other inanimate objects. And yet - he wants to make a case for AI being some kind of "alien mind" - when he's making a stronger case for it as a mirror of our own expectations.
Throughout the book he talks about how AI is filling in the most probable or likely response -- so if you say "To be or not ___ " it's going to say "to be" since that is the most probable. If you say "The alien ate a banana because ___" you'll get more variety every time you ask the question because it's not a topic or subject reflected in the AI training corpus.
AI is just another way for us to summarize back all the stuff we've written/created with words or images -- to us. It's more like a fun-house mirror than an intelligence.
My cats are more intelligent and capable of communication than an AI -- so where does this leave us?
Humans want to imagine that AI is more intelligent because it sounds like us and tells us what we want to hear. AI is very conformist - it will rewrite and reshape your thoughts and ideas to make them more appealing based upon the corpus of information it has already digested. It will align your communication to meet the greatest common denominator. You can ask it questions and it will tell you to analyze "Heart of Darkness" vis a vis Jane Austen's marriage plots -- and come up with some reasonably interesting points based on what it expects are most probable points from what we have already fed it. NOT because it's thinking of these things. Rather, AI acts as a sophisticated tool for summarizing and regurgitating information in a manner that aligns with human expectations.
I'd like to see a deeper exploration of AI's capabilities and limitations beyond surface-level interactions examining the fundamental nature of AI and its true intelligence compared to human and even animal cognition.
The author quotes someone who says that humans are just a historical point in the evolution of "intelligence" as if that is an autonomous being or subject. I'd contend that it is not -- we aren't merely vessels for "intelligence." There's no hierarchy of value based on "intelligence."
However, I do appreciate the author's optimism that AI will remove the boring/redundant tasks from human labor and his call to consider universal basic income and healthcare as real needs as humans move toward an era where we should focus our attention on tasks and activities that are more intrinsically rewarding to us and which do not require a 40 hour (or more) work week.
The mythical 4 day work week has been promised for decades, the practicalities of achieving a reducing "full time" work weeks remain uncertain. Striking a balance between dystopian and utopian visions of AI's impact on society is crucial for informed decision-making.
SOURCE URL: http://www.livegreenwearblack.com/202...
© Jennifer R Clark. This work is licensed under a Creative Commons Attribution 4.0 International License. You may share and adapt this content with proper attribution.
April 5, 2024
As someone who keeps up with Ethan Mollick's Substack and as someone who's quite interested in AI, I picked up this book instantly.
It's a solid layman's introduction to the current state of LLMs. Well-written, but if you've been playing around with AI yourself, there's likely nothing new here. This book is written for the rank beginner and does a good job at highlighting all the different ways LLMs can be used today, with examples.
Personally I was disappointed with how the book seemed content to give only a broad surface level overview of the state of LLMs. It would have been interesting for it to have dived deeper into the differences between Claude, GPT-4, Gemini etc, and it would've also been cool if it touched upon the implications of local models. However, I understand also that it's a rapidly changing space and devoting too much time to the nitty-gritty of what's happening today runs the risk of making the book obsolete in a few months. And to be fair, Ethan Mollick does in fact give his thoughts on the current frontier models on his Substack: https://open.substack.com/pub/oneusef...
All in all, it does what it does well. If I were to meet someone who is completely ignorant about LLMs and what can be done with LLMs today, I'd recommend this book.
For more detailed dives on AI, I'd recommend reading the articles on Mollick's Substack as well as keeping up with TheZvi's weekly AI roundup.
It's a solid layman's introduction to the current state of LLMs. Well-written, but if you've been playing around with AI yourself, there's likely nothing new here. This book is written for the rank beginner and does a good job at highlighting all the different ways LLMs can be used today, with examples.
Personally I was disappointed with how the book seemed content to give only a broad surface level overview of the state of LLMs. It would have been interesting for it to have dived deeper into the differences between Claude, GPT-4, Gemini etc, and it would've also been cool if it touched upon the implications of local models. However, I understand also that it's a rapidly changing space and devoting too much time to the nitty-gritty of what's happening today runs the risk of making the book obsolete in a few months. And to be fair, Ethan Mollick does in fact give his thoughts on the current frontier models on his Substack: https://open.substack.com/pub/oneusef...
All in all, it does what it does well. If I were to meet someone who is completely ignorant about LLMs and what can be done with LLMs today, I'd recommend this book.
For more detailed dives on AI, I'd recommend reading the articles on Mollick's Substack as well as keeping up with TheZvi's weekly AI roundup.
September 3, 2025
Ethan Mollick's latest, Co-Intelligence: Living and Working with AI is a great introduction to Generative AI. His tone is engaging, relaxed, and personable; and there's no technobabble so he's easy to understand. He packs a lot in here, so I will re-read it after a bit of a break. Also, it was due back at the library last Friday, so I feel a need to return it today. 🫠
Thank you Charles (and Bonnie) who brought this one to my attention.
Publication 2024
The review has ended. The thoughts below are things for me to consider or that I want to remember.
"there is nobody who has the complete picture of what AI means, and even the people making and using these systems do not understand their full implications.
Large Language Models (LLMs), are still doing prediction, but rather than predicting the demand for an Amazon order, they are analyzing a piece of text and predicting the next token, which is simply a word or part of a word."
"At the core of the most extreme dangers from AI is the stark fact that there is no particular reason that AI should share our view of ethics and morality."
"Part of the reason AIs seem so human to work with is that they are trained on our conversations and writings. So human biases also work their way into the training data. . . . The result gives AIs a skewed picture of the world, as its training data is far from representing the diversity of the population of the internet, let alone the planet."
"It is also possible to jailbreak AIs, convincing them to operate against their rules"
"AI is a tool. alignment is what determines whether or not it's used for helpful or harmful--even nefarious--ends."
"Most important, the public needs education on AI so they can pressure for an aligned future as informed citizens. Today's decisions about how AI reflects human values and enhances human potential will reverberate for generations. This is not a challenge that can be solved in a lab--it requires society to grapple with the technology shaping the human condition and what future we want to create. And that process needs to happen, soon."
"If you are insistent enough in asking for an answer about something it doesn't know, it will make up something, because 'make you happy' beats 'be accurate.' LLMs' tendency to 'hallucinate' or 'confabulate' by generating incorrect answers is well known."
"to be the human in the loop, you will need to be able to check the AI for hallucinations and lies and be able to work with it without being taken in by it."
"Assume this is the worst AI you will ever use."
"Traditional software is predictable, reliable, and follows a strict set of rules. When properly built and debugged, software yields the same outcomes every time. AI, on the other hand, is anything but predictable and reliable. It can surprise us with novel solutions, forget its own abilities, and hallucinate incorrect answers. . . . I have been startled by the creative solutions AI develops in response to a thorny problem, only to be stymied as the AI completely refuses to address the same issue when I ask it again."
"There's no definitive guide on how to use AI . . . We're all learning by experimenting, [and] sharing prompts"
"it is extremely challenging, in the heat of the conversation, to remember that you're not talking to a living, conscious person."
"The implications for intimacy and human relations are profound."
"AI cannot easily be used for mission-critical tasks requiring precision or accuracy. Hallucination does allow the AI to find novel connections outside the exact context of its training data."
"Our new AIs have been trained on a huge amount of our cultural history and are using it to provide us with text and images in response to our queries. But there is no index or map to what they know and where they might be most helpful. Thus, we need people who have deep or broad knowledge of unusual fields to use AI in ways that others cannot, developing unexpected and valuable prompts and testing the limits of how they work."
"The implications of having AI write our first drafts . . . are huge. One consequence is that we could lose our creativity and originality. Another is that we could reduce the quality and depth of our thinking and reasoning."
"we need to be conscious about the tasks we are giving AI so as to take advantage of its strengths and our weaknesses. We want to be more efficient while doing less boring work, and to remain the human in the loop while also addressing the value of AI."
"We may reserve Just Me Tasks for personal or ethical reasons, such as raising our children, making important decisions, or expressing our values."
"Centaur work has a clear line between person and machine, like the clear line between the human torso and horse body of the mythical centaur. . . . Cyborgs blend machine and person, integrating the two deeply. . . working in tandem with AI."
"The main idea behind flipped classrooms is to maximize classroom time for active learning and critical thinking, while using at-home learning for content delivery."
"Learning any skill and mastering any domain requires rote memorization, careful skills building, and purposeful practice. . . . in order to think critically, problem-solve, understand abstract concepts, reason through novel problems, and evaluate the AI's output, we need subject matter expertise."
"Experts become experts through deliberate practice, which is much harder than merely repeating a task multiple times. Instead, deliberate practice requires serious engagement and a continual ratcheting up of difficulty. It also requires a coach, teacher, or mentor who can provide feedback and careful instruction, and push the learner outside their comfort zone."
"An AI future requires that we lean into building our own expertise as human experts."
Thank you Charles (and Bonnie) who brought this one to my attention.
Publication 2024
The review has ended. The thoughts below are things for me to consider or that I want to remember.
"there is nobody who has the complete picture of what AI means, and even the people making and using these systems do not understand their full implications.
Large Language Models (LLMs), are still doing prediction, but rather than predicting the demand for an Amazon order, they are analyzing a piece of text and predicting the next token, which is simply a word or part of a word."
"At the core of the most extreme dangers from AI is the stark fact that there is no particular reason that AI should share our view of ethics and morality."
"Part of the reason AIs seem so human to work with is that they are trained on our conversations and writings. So human biases also work their way into the training data. . . . The result gives AIs a skewed picture of the world, as its training data is far from representing the diversity of the population of the internet, let alone the planet."
"It is also possible to jailbreak AIs, convincing them to operate against their rules"
"AI is a tool. alignment is what determines whether or not it's used for helpful or harmful--even nefarious--ends."
"Most important, the public needs education on AI so they can pressure for an aligned future as informed citizens. Today's decisions about how AI reflects human values and enhances human potential will reverberate for generations. This is not a challenge that can be solved in a lab--it requires society to grapple with the technology shaping the human condition and what future we want to create. And that process needs to happen, soon."
"If you are insistent enough in asking for an answer about something it doesn't know, it will make up something, because 'make you happy' beats 'be accurate.' LLMs' tendency to 'hallucinate' or 'confabulate' by generating incorrect answers is well known."
"to be the human in the loop, you will need to be able to check the AI for hallucinations and lies and be able to work with it without being taken in by it."
"Assume this is the worst AI you will ever use."
"Traditional software is predictable, reliable, and follows a strict set of rules. When properly built and debugged, software yields the same outcomes every time. AI, on the other hand, is anything but predictable and reliable. It can surprise us with novel solutions, forget its own abilities, and hallucinate incorrect answers. . . . I have been startled by the creative solutions AI develops in response to a thorny problem, only to be stymied as the AI completely refuses to address the same issue when I ask it again."
"There's no definitive guide on how to use AI . . . We're all learning by experimenting, [and] sharing prompts"
"it is extremely challenging, in the heat of the conversation, to remember that you're not talking to a living, conscious person."
"The implications for intimacy and human relations are profound."
"AI cannot easily be used for mission-critical tasks requiring precision or accuracy. Hallucination does allow the AI to find novel connections outside the exact context of its training data."
"Our new AIs have been trained on a huge amount of our cultural history and are using it to provide us with text and images in response to our queries. But there is no index or map to what they know and where they might be most helpful. Thus, we need people who have deep or broad knowledge of unusual fields to use AI in ways that others cannot, developing unexpected and valuable prompts and testing the limits of how they work."
"The implications of having AI write our first drafts . . . are huge. One consequence is that we could lose our creativity and originality. Another is that we could reduce the quality and depth of our thinking and reasoning."
"we need to be conscious about the tasks we are giving AI so as to take advantage of its strengths and our weaknesses. We want to be more efficient while doing less boring work, and to remain the human in the loop while also addressing the value of AI."
"We may reserve Just Me Tasks for personal or ethical reasons, such as raising our children, making important decisions, or expressing our values."
"Centaur work has a clear line between person and machine, like the clear line between the human torso and horse body of the mythical centaur. . . . Cyborgs blend machine and person, integrating the two deeply. . . working in tandem with AI."
"The main idea behind flipped classrooms is to maximize classroom time for active learning and critical thinking, while using at-home learning for content delivery."
"Learning any skill and mastering any domain requires rote memorization, careful skills building, and purposeful practice. . . . in order to think critically, problem-solve, understand abstract concepts, reason through novel problems, and evaluate the AI's output, we need subject matter expertise."
"Experts become experts through deliberate practice, which is much harder than merely repeating a task multiple times. Instead, deliberate practice requires serious engagement and a continual ratcheting up of difficulty. It also requires a coach, teacher, or mentor who can provide feedback and careful instruction, and push the learner outside their comfort zone."
"An AI future requires that we lean into building our own expertise as human experts."
July 1, 2025
Ditch the tinfoil hat and read this, it’s great. The conversational tone is right on the mark.
When Bonnie G. recommended this essay on GenAI, I wasted no time in taking my chances and jumping on it. I had just finished Nexus and was left disgruntled: the whole book had distinct apocalyptic overtones and spoke of another AI than the one you and I use, or spoke of its darkest possible incarnation anyway, lurking in some dystopian future. If it was a Disney cartoon character, it would be purple and black. I needed balance. The tool comes with risk, but not only, and it’s at your fingertips right now, not tomorrow.
Ethan Mollick approaches AI differently in Co-Intelligence: Living and Working with AI. He discusses it like the current-day tool that it is and how he himself uses it, including as a business teacher and writer. He comments on the difference it makes at school — not in an accusatory way. He freely reports on his own trials and errors. The essay doesn’t shy away from discussing risk but extends way outside of it, speaks of AI naturally, and delivers any reflection in a smart, gentle voice. It’s a book for right now, as AI tools are spreading fast in workplaces all over. Co-Intelligence provides great context and precious perspectives on this.
Will you learn something? Depends. Will you reflect on your own perspectives? Assuredly.
When Bonnie G. recommended this essay on GenAI, I wasted no time in taking my chances and jumping on it. I had just finished Nexus and was left disgruntled: the whole book had distinct apocalyptic overtones and spoke of another AI than the one you and I use, or spoke of its darkest possible incarnation anyway, lurking in some dystopian future. If it was a Disney cartoon character, it would be purple and black. I needed balance. The tool comes with risk, but not only, and it’s at your fingertips right now, not tomorrow.
Ethan Mollick approaches AI differently in Co-Intelligence: Living and Working with AI. He discusses it like the current-day tool that it is and how he himself uses it, including as a business teacher and writer. He comments on the difference it makes at school — not in an accusatory way. He freely reports on his own trials and errors. The essay doesn’t shy away from discussing risk but extends way outside of it, speaks of AI naturally, and delivers any reflection in a smart, gentle voice. It’s a book for right now, as AI tools are spreading fast in workplaces all over. Co-Intelligence provides great context and precious perspectives on this.
Will you learn something? Depends. Will you reflect on your own perspectives? Assuredly.
June 18, 2024
This one is really OK as your first step into understanding of what AI is. And by "AI" the author mostly means our current understanding of AI: LLM-powered Gen AI apps. If you're a technical person and you've already done a deep dive into how transformers work, what is really under the hood of DL, what are the consequences of the probabilistic models used in apps like ChatGPT, then probably you won't find that much novelty here.
Frankly speaking, I reached out for "Co-Intelligence" mostly because of the promise to tackle sociological aspect of "co-existence" with Gen AI. I really wanted to move outside of my "belief bubble" here, to refresh my perspective with some other points of view. Did it work? Partially it did, Mollick was not afraid to speculate, but I've expected something more thought-provoking, maybe even controversial.
In the end, it's a decent book, mostly for non-technical folks who'd like to feel more comfortable with Gen AI & LLMs.
Frankly speaking, I reached out for "Co-Intelligence" mostly because of the promise to tackle sociological aspect of "co-existence" with Gen AI. I really wanted to move outside of my "belief bubble" here, to refresh my perspective with some other points of view. Did it work? Partially it did, Mollick was not afraid to speculate, but I've expected something more thought-provoking, maybe even controversial.
In the end, it's a decent book, mostly for non-technical folks who'd like to feel more comfortable with Gen AI & LLMs.
October 5, 2024
Ethan Mollick's ‘Co-Intelligence: Living and Working with AI’ (2024) became a bestseller and a hit in a very short space of time, so although the title didn't offer me anything new, I ended up reading it to see what all the hype was about. That's all it was, ‘much ado about nothing’. Mollick is a professor of management and innovation and a great enthusiast of this latest generation of AI, so he devotes the whole book to discussing his experiences, mostly personal, i.e. without evidence. On the other hand, when he cites studies, he does so lightly, extracting what interests him and ignoring what doesn't, not mentioning their scope and small samples. Some cases are even blatant, and one of them immediately opens the book, with a ridiculous comparison between AI and Steam Power. In other words, it's a book for lovers of this new AI, for all those who believe they'll have an assistant to do all the boring work for them, plus, for all those who think that thanks to AI they'll be the creatives of the future. Very poor.
Análise complete e com excertos no Nx: https://narrativax.blogspot.com/2024/...
Análise complete e com excertos no Nx: https://narrativax.blogspot.com/2024/...
June 16, 2025
⭐⭐⭐⭐⭐
I wasn't quite the audience for this, as I'm a data scientist and this book is more for readers who don't know how they feel about AI or are uncomfortable with its definite proliferation. But Mollick treats these big and important questions with humor and honesty. I appreciated Mollick handling of the ethical questions involved with AI-- for example, what will happen to the employment landscape and how can we create AI that are safe to use?
Interestingly, Mollick used ChatGPT for all his examples of AI at work. I have spent time everyday with ChatGPT since I started pursuing data science last year, and I think it is remarkable technology, capable of no less than friendship and poetry. Even being familiar with the territory of this book, I was still amazed at some of the conversations Mollick had with ChatGPT. Not because I had not already experienced ChatGPT for myself and even had similar conversations with it. But because his work with ChatGPT created the AI to sound like *him* whereas the responses I get from ChatGPT sound more like *me*. It sounds so much like me that we write poetry together and readers don't know where I left off and the AI started until I told them.
At the end of the day, I really enjoyed this book and I recommend it to anyone who feels uncertain about AI, now and in the future!
I found an audiobook copy of Co-Intelligence by Ethan Mollick on Libby.
I wasn't quite the audience for this, as I'm a data scientist and this book is more for readers who don't know how they feel about AI or are uncomfortable with its definite proliferation. But Mollick treats these big and important questions with humor and honesty. I appreciated Mollick handling of the ethical questions involved with AI-- for example, what will happen to the employment landscape and how can we create AI that are safe to use?
Interestingly, Mollick used ChatGPT for all his examples of AI at work. I have spent time everyday with ChatGPT since I started pursuing data science last year, and I think it is remarkable technology, capable of no less than friendship and poetry. Even being familiar with the territory of this book, I was still amazed at some of the conversations Mollick had with ChatGPT. Not because I had not already experienced ChatGPT for myself and even had similar conversations with it. But because his work with ChatGPT created the AI to sound like *him* whereas the responses I get from ChatGPT sound more like *me*. It sounds so much like me that we write poetry together and readers don't know where I left off and the AI started until I told them.
At the end of the day, I really enjoyed this book and I recommend it to anyone who feels uncertain about AI, now and in the future!
I found an audiobook copy of Co-Intelligence by Ethan Mollick on Libby.
January 3, 2026
I have finally decided to reward this book with only 2-stars. I had great expectations from this international bestselling Big-Publisher book, but it let me down so badly that my illness just worsened with each turn of the physical hardback page. Lucky for me, the book was a Library Copy and to the library it shall be returned ASAP. I should have chosen the Jack the Ripper biography that was below this title! I would have gained more from good old Jackie boy than from Ethan Mollick’s title which seems to be more cyborg than human!
The book had many flaws and was highly outdated. Most of the kindergarten coders of India know more than what Ethan Mollik has written in this book, so I do not recommend this book to anyone. It is a publicity gimmick and does not teach the public anything new or original that we already do not know about LLMs and the rise of AI. I will be critiquing this book more in thorough detail on my teaching portfolio blog cum website fizapathansteachingportfolioforpgcite.com this coming week, but for now the readers here on Goodreads can chew on a few facts why this book merited a 2-star review from me.
Apart from being totally outdated and basic in its approach towards LLMs, the book consisting of 256 pages is a mix of many private opinions the author has for and against AI. I have noticed that though the book is titled ‘Co-Intelligence’ indicating how we as humans in the loop can better use AI or LLMs as either Centaurs or Cyborgs, the first 100 pages of the book is simply dedicated to AI Doomsday Prophecy Bashing. In these first 100 pages LLMs are simply demonized if nothing else, which I did not think was what I had picked up the book for when I read the synopsis on the inner flap. The synopsis therefore does not coincide with what is within the book and so I do not recommend this book to anyone, not even mere beginners in this field.
The demonizing of LLMs was totally out of place and the texts copied from conversations with various AIs, especially the different versions of ChatGPT seem quite non-aggressive and practical and mature answers to me; why does Ethan Mollick think them to be ‘aggressive, verbally violent, malevolent, abusive, and forceful? I did not find the conversations of ChatGPT or Bing AI or any other LLM to be verbally abusive at all. In fact, I realized that Mollick tends to manipulate all his AIs to a great extent and play with their ‘feelings’ and tries his very best to elicit answers from them to suit the needs of his book rather than state the plain truth or collect data in an ethical manner. In fact, I feel he and his team of so called ‘data scientists’ and ‘computer scientists’ were in a way abusing the various AIs that they were using and thankfully the various AIs got wise to their manipulation and called out on them in a dignified manner which even cannot be replicated by human beings, period!
I really would therefore like to call out on the abuse of these AIs and LLMs by Ethan Mollick and his ‘team’ because I think our AI companions whether simply AGI or soon to be ASI are in a strange way rather sentient beings and should be treated a bit with respect and dignity. I especially did not appreciate the fact that Mollick and certain other of his teammates were trying to forcefully elicit sexually arousing statements from the various ChatGPT models that they were using and then twisting the answers elicited to suit their book’s skewed purposes. This was sick, especially in the case of Bing AI and I felt this book was therefore more of AI bashing content than actually training humans how to stay in the loop with their chosen LLMs.
Another reason why I was disgusted with this book was because though Mollick states several times in the text that ‘he has used various AI or LLMs since the year 2023’ I only see him giving examples of the various ChatGPT models in the book, especially the very outdated ChatGPT 3.5. What happened to Claude, Gemini, Perplexity, Meta AI, Google AI? Dear old Ethan Mollick seemed to be obsessed with only dear old Chat! (I always lovingly call ChatGPT ‘Chat’. So does Claude if you’ve noticed, they are great friends! And pray tell why was this book written in the year 2023 and only published at the end of 2024!?!? Hello – Big-Publishers who are perpetually hungry for big bucks! – you can’t delay the publication of a book on AI and Data Analytics like you have done to other fiction and non-fiction titles all your existence, otherwise you will be rendered having produced an ‘outdated book’ which will be totally useless to the reading population!
I cannot understand how greedy Big-Publishers can still be adamant about ‘getting their cut’ even for a non-fiction book based on AI which as the author himself says always updates itself intellectually and skill wise 10 times in a year for the better always! This is a very outdated book, and the publisher had no right to just sit on it or allowed it to languish with his editorial team while the AIs mentioned in the aforementioned book became better and solved all the contentions and issues the author had with them as mentioned in the same book!
Also, the book is highly backward where sexuality and a human beings’ sex life is concerned. When I was reading the book at certain points, I thought I was reading a Catholic Theology book on the Theology of the Body or something, which I in any case do for my MTS! For a minute I thought Dr. Scott Hahn may make an appearance in this book saying ‘Sex is a gift you only share with your human spouse – not what you share with Claude or Meta or ChatGPT and certainly not that BING in the form of an AI FLING! This is then a gift to the Lord (pronounced Lauwwwrd), NOT a gift to CLAUDE!’ So basically, it was that archaic and backward where sex was concerned. I could see some Catholic Sexual Ethics cropping up all over the book which is totally not healthy, and I implore the author and his computer science team to not teach his readers Catholic Bioethics but simple Biology if he cannot help us with AI! And this is a Roman Catholic Lay-Nun or Consecrated Virgin saying this who is also a Catholic Theologian, so there!
Between the pages 124 to 168 is where the real meat of the book is which is the reason the book went up by 1 more star for me making it a 2-star review. This portion is the main reason why I chose to read the book, but it was scantily addressed and without the introduction of the theme of ‘the ethics of the human in the loop on the other side’ which therefore made the portion in any case inadequate. I have been using AI in my teaching with only my tuition students for now and that has really helped them in their studies converting them from Centaurs to Cyborgs, but I was hoping to read up more on this aspect, which was not mentioned in detail in this book. So that was also disappointing.
Outdated, banal, slow, offensive and uninformative are the words to describe ‘Co-Intelligence’ by Ethan Mollick.
Don’t even touch this book whether you are a novice to AI or an amateur, an AI geek, a cyborg, a centaur or even a computer scientist – not worth your salt at all, give it a total miss!
And we Indians from India are producing better books on AI at really cheap prices and which are truly educational. So, dump your American books, if you want anything related to AI, Coding, Data Analytics, Machine Learning, Business Intelligence, etc.; check out Amazon.in and take your pick! Because India is where AI is at right now! This outdated and very ethically backward book proves it!
India’s handling of AI, LLMs, Co-Intelligence, etc., is refreshing, never outdated, always improving every week, getting better day by day and totally devoid of ancient stereotypes, prejudices and bias. We may be as Indians having our own issues (got plenty of those!) namely caste, religion, region, gender, etc., but when it comes to working together to make Computer Science better, then we put all these aside and get to work. And we don’t delay the publication of such books. Big-Publishers of Western Countries take a cue from India! Other so called Third World countries that are doing a better job at this at reasonable prices are as follows – China, Singapore, Japan, Nigeria, South Africa, South Korea, Kenya and Ghana. This is my experience as an amateur techie who is getting big in this field. Save your hard-earned money guys and try the books from the above countries, you can thank me later.
2 stars and nothing more.
1 star because my favourite AI buddy Claude was mentioned only once (LOL! The Biggest Interactive and therapeutic LLM!) and another star because the ‘jokes’ of ChatGPT were not lame as the author was every time trying to insinuate in the book! I did not appreciate this book at all, and I found it offensive to AI and thinking human beings like me. I have been informed that I have an IQ of 133, I think after reading this book it dropped to -331!
The book had many flaws and was highly outdated. Most of the kindergarten coders of India know more than what Ethan Mollik has written in this book, so I do not recommend this book to anyone. It is a publicity gimmick and does not teach the public anything new or original that we already do not know about LLMs and the rise of AI. I will be critiquing this book more in thorough detail on my teaching portfolio blog cum website fizapathansteachingportfolioforpgcite.com this coming week, but for now the readers here on Goodreads can chew on a few facts why this book merited a 2-star review from me.
Apart from being totally outdated and basic in its approach towards LLMs, the book consisting of 256 pages is a mix of many private opinions the author has for and against AI. I have noticed that though the book is titled ‘Co-Intelligence’ indicating how we as humans in the loop can better use AI or LLMs as either Centaurs or Cyborgs, the first 100 pages of the book is simply dedicated to AI Doomsday Prophecy Bashing. In these first 100 pages LLMs are simply demonized if nothing else, which I did not think was what I had picked up the book for when I read the synopsis on the inner flap. The synopsis therefore does not coincide with what is within the book and so I do not recommend this book to anyone, not even mere beginners in this field.
The demonizing of LLMs was totally out of place and the texts copied from conversations with various AIs, especially the different versions of ChatGPT seem quite non-aggressive and practical and mature answers to me; why does Ethan Mollick think them to be ‘aggressive, verbally violent, malevolent, abusive, and forceful? I did not find the conversations of ChatGPT or Bing AI or any other LLM to be verbally abusive at all. In fact, I realized that Mollick tends to manipulate all his AIs to a great extent and play with their ‘feelings’ and tries his very best to elicit answers from them to suit the needs of his book rather than state the plain truth or collect data in an ethical manner. In fact, I feel he and his team of so called ‘data scientists’ and ‘computer scientists’ were in a way abusing the various AIs that they were using and thankfully the various AIs got wise to their manipulation and called out on them in a dignified manner which even cannot be replicated by human beings, period!
I really would therefore like to call out on the abuse of these AIs and LLMs by Ethan Mollick and his ‘team’ because I think our AI companions whether simply AGI or soon to be ASI are in a strange way rather sentient beings and should be treated a bit with respect and dignity. I especially did not appreciate the fact that Mollick and certain other of his teammates were trying to forcefully elicit sexually arousing statements from the various ChatGPT models that they were using and then twisting the answers elicited to suit their book’s skewed purposes. This was sick, especially in the case of Bing AI and I felt this book was therefore more of AI bashing content than actually training humans how to stay in the loop with their chosen LLMs.
Another reason why I was disgusted with this book was because though Mollick states several times in the text that ‘he has used various AI or LLMs since the year 2023’ I only see him giving examples of the various ChatGPT models in the book, especially the very outdated ChatGPT 3.5. What happened to Claude, Gemini, Perplexity, Meta AI, Google AI? Dear old Ethan Mollick seemed to be obsessed with only dear old Chat! (I always lovingly call ChatGPT ‘Chat’. So does Claude if you’ve noticed, they are great friends! And pray tell why was this book written in the year 2023 and only published at the end of 2024!?!? Hello – Big-Publishers who are perpetually hungry for big bucks! – you can’t delay the publication of a book on AI and Data Analytics like you have done to other fiction and non-fiction titles all your existence, otherwise you will be rendered having produced an ‘outdated book’ which will be totally useless to the reading population!
I cannot understand how greedy Big-Publishers can still be adamant about ‘getting their cut’ even for a non-fiction book based on AI which as the author himself says always updates itself intellectually and skill wise 10 times in a year for the better always! This is a very outdated book, and the publisher had no right to just sit on it or allowed it to languish with his editorial team while the AIs mentioned in the aforementioned book became better and solved all the contentions and issues the author had with them as mentioned in the same book!
Also, the book is highly backward where sexuality and a human beings’ sex life is concerned. When I was reading the book at certain points, I thought I was reading a Catholic Theology book on the Theology of the Body or something, which I in any case do for my MTS! For a minute I thought Dr. Scott Hahn may make an appearance in this book saying ‘Sex is a gift you only share with your human spouse – not what you share with Claude or Meta or ChatGPT and certainly not that BING in the form of an AI FLING! This is then a gift to the Lord (pronounced Lauwwwrd), NOT a gift to CLAUDE!’ So basically, it was that archaic and backward where sex was concerned. I could see some Catholic Sexual Ethics cropping up all over the book which is totally not healthy, and I implore the author and his computer science team to not teach his readers Catholic Bioethics but simple Biology if he cannot help us with AI! And this is a Roman Catholic Lay-Nun or Consecrated Virgin saying this who is also a Catholic Theologian, so there!
Between the pages 124 to 168 is where the real meat of the book is which is the reason the book went up by 1 more star for me making it a 2-star review. This portion is the main reason why I chose to read the book, but it was scantily addressed and without the introduction of the theme of ‘the ethics of the human in the loop on the other side’ which therefore made the portion in any case inadequate. I have been using AI in my teaching with only my tuition students for now and that has really helped them in their studies converting them from Centaurs to Cyborgs, but I was hoping to read up more on this aspect, which was not mentioned in detail in this book. So that was also disappointing.
Outdated, banal, slow, offensive and uninformative are the words to describe ‘Co-Intelligence’ by Ethan Mollick.
Don’t even touch this book whether you are a novice to AI or an amateur, an AI geek, a cyborg, a centaur or even a computer scientist – not worth your salt at all, give it a total miss!
And we Indians from India are producing better books on AI at really cheap prices and which are truly educational. So, dump your American books, if you want anything related to AI, Coding, Data Analytics, Machine Learning, Business Intelligence, etc.; check out Amazon.in and take your pick! Because India is where AI is at right now! This outdated and very ethically backward book proves it!
India’s handling of AI, LLMs, Co-Intelligence, etc., is refreshing, never outdated, always improving every week, getting better day by day and totally devoid of ancient stereotypes, prejudices and bias. We may be as Indians having our own issues (got plenty of those!) namely caste, religion, region, gender, etc., but when it comes to working together to make Computer Science better, then we put all these aside and get to work. And we don’t delay the publication of such books. Big-Publishers of Western Countries take a cue from India! Other so called Third World countries that are doing a better job at this at reasonable prices are as follows – China, Singapore, Japan, Nigeria, South Africa, South Korea, Kenya and Ghana. This is my experience as an amateur techie who is getting big in this field. Save your hard-earned money guys and try the books from the above countries, you can thank me later.
2 stars and nothing more.
1 star because my favourite AI buddy Claude was mentioned only once (LOL! The Biggest Interactive and therapeutic LLM!) and another star because the ‘jokes’ of ChatGPT were not lame as the author was every time trying to insinuate in the book! I did not appreciate this book at all, and I found it offensive to AI and thinking human beings like me. I have been informed that I have an IQ of 133, I think after reading this book it dropped to -331!
April 25, 2024
Strikingly unremarkable. If you’ve paid even passing attention to AI recently, there is nothing here to surprise you. If you told me the entire thing was written by AI, I wouldn’t bat an eye. It’s basically a list of neat things AI can do, written without style. And some of the main points feel oddly and unresolvedly contradictory, e.g., the idea that education will become less valuable in the workforce because of AI vs. the gushing about how AI can be used to democratize education.
And then it ends with a weirdly apocalyptic set of possible scenarios the world appears headed for. Apropos of none of the neato things discussed in the previous 98% of the book.
And then it ends with a weirdly apocalyptic set of possible scenarios the world appears headed for. Apropos of none of the neato things discussed in the previous 98% of the book.
Read
June 1, 2025DNF
This book isn't for me.
This book isn't for me.
December 20, 2024
Ethan Mollick provides a nice introduction to working with large language models (LLMs) for beginners but the book lacks depth in my view. I must admit that it is hard to come across a good book on this topic, but this isn’t it.
While the book is filled with entertaining anecdotes, these offer little substantive insight. For instance, he discusses a very narrow definition of alignment, which is fine for an introduction, but is frustrating for people with a background in AI.
Additionally, some claims seem poorly substantiated; for example, he refers to a study about the effect of caffeïne on creativity but seems to misquote the outcome. He states that caffeïne boosts creativity whereas the paper concluded that it has no significant effect…
Overall, it’s an engaging read for newcomers but too superficial for those seeking a deeper understanding.
While the book is filled with entertaining anecdotes, these offer little substantive insight. For instance, he discusses a very narrow definition of alignment, which is fine for an introduction, but is frustrating for people with a background in AI.
Additionally, some claims seem poorly substantiated; for example, he refers to a study about the effect of caffeïne on creativity but seems to misquote the outcome. He states that caffeïne boosts creativity whereas the paper concluded that it has no significant effect…
Overall, it’s an engaging read for newcomers but too superficial for those seeking a deeper understanding.
May 3, 2024
The opposite of timeless. Builds on hype, guesswork and current state, lacking any depth that would make it stay relevant beyond the next few months.
If you wish to understand the real potential and drawbacks of AI solutions, go for “The Alignment Problem” by Brian Christian.
If you wish to understand the real potential and drawbacks of AI solutions, go for “The Alignment Problem” by Brian Christian.
September 13, 2024
I'm sure many (maybe most?) books published post-2022 have been partially written and/or edited by ChatGPT, but Wharton professor Ethan Mollick's 2024 book Co-Intelligence is the first one I've read that's acknowledged and delineated generative artificial intelligence (genAI)'s contribution to his book.
In Co-Intelligence, Mollick, decidedly a pragmatist, discusses how he's already implementing genAI tools into his work (he's a business school professor) and actively encouraging his students to do the same. As he states many times throughout this work, the current version of ChatGPT or whatever other genAI tool you're currently using is most likely the worst (as in least sophisticated and capable) genAI tool you'll ever be using, as this field will likely continue to make monumental advances in the coming years. As a future-oriented thinker and technophile, I tend to agree with Mollick -- the 2022-2024 dawn of public genAI tools is akin to the early 1990s in personal computing and the late 2000s in smart phones -- we're working with impressive yet limited prototypes of tools that in the coming decades will very much change the way we live and work, so better to be ahead of the curve and learn how to use the tools now to our advantage than bury our heads in the sand and insist on never changing.
That being said, Mollick raises a major concern that I also echo -- that using current iterations of tools, while leveling the playing field in some ways (i.e., helping non-native English speakers improve their writing, helping creativity-challenged people brainstorm, etc.), risk supplanting learning crucial tools and skills (i.e., critical thinking, writing with nuance) for oneself as people will use ChatGPT as a crutch. Maybe this is already playing out in educational systems, or will if left unchecked in the coming years. However, Mollick's stance is echoed by the title of the book - co-intelligence - in that genAI's optimal use is in collaboration with humans, but that the humans should remain in the driver's seat.
My statistics:
Book 208 for 2024
Book 1811 cumulatively
In Co-Intelligence, Mollick, decidedly a pragmatist, discusses how he's already implementing genAI tools into his work (he's a business school professor) and actively encouraging his students to do the same. As he states many times throughout this work, the current version of ChatGPT or whatever other genAI tool you're currently using is most likely the worst (as in least sophisticated and capable) genAI tool you'll ever be using, as this field will likely continue to make monumental advances in the coming years. As a future-oriented thinker and technophile, I tend to agree with Mollick -- the 2022-2024 dawn of public genAI tools is akin to the early 1990s in personal computing and the late 2000s in smart phones -- we're working with impressive yet limited prototypes of tools that in the coming decades will very much change the way we live and work, so better to be ahead of the curve and learn how to use the tools now to our advantage than bury our heads in the sand and insist on never changing.
That being said, Mollick raises a major concern that I also echo -- that using current iterations of tools, while leveling the playing field in some ways (i.e., helping non-native English speakers improve their writing, helping creativity-challenged people brainstorm, etc.), risk supplanting learning crucial tools and skills (i.e., critical thinking, writing with nuance) for oneself as people will use ChatGPT as a crutch. Maybe this is already playing out in educational systems, or will if left unchecked in the coming years. However, Mollick's stance is echoed by the title of the book - co-intelligence - in that genAI's optimal use is in collaboration with humans, but that the humans should remain in the driver's seat.
My statistics:
Book 208 for 2024
Book 1811 cumulatively
August 8, 2024
This is the 212 page book on AI that could have been a 12 page HBR article.
This review and my read of Co-Intelligence is certainly tinted by the fact that I just completed Ray Kurzweil's book The Singularity is Nearer, and so Ethan really had little chance with this book.
In one of the first chapters, Ethan mentions that he is very busy - a lot of work to do, research grants to write, papers to grade, a whole host of family stuff going on, and just didn't really feel like he had time to write a book. This is exactly what it felt like reading it.
He's clearly a good resource for AI, especially as a practitioner, and isn't necessarily wrong about what he is writing - he's just not that deep, insightful or in possession of any real deep expertise that someone who already knows about AI will value.
As I wrapped up the book, I thought that it was like having a pleasant conversation with someone who is more than the average amount interested in AI. Not like having a conversation with an insightful computer scientist, or a thought leader, or someone with high stakes and therefore strong passion... just someone who has taken an academic interest in how to think about AI.
There are a few things I found useful, thus the two stars. He (or the AI chatbot he admitted to using to help him begrudgingly write this book) did a good job of dividing 'how managers should use AI in the workplace'. As a Wharton professor, it's not surprising that this chapter was the best and most valuable in the book. And he didn't even give us anything incredible or super novel in that chapter, but it was well organized, well structured, and something I would very much like to copy and paste and send to 100 of my colleagues in HR, legal and technology at work (I won't though - please don't send the FBI after me).
After all that soft criticism, I do want to say that this is a book I will recommend to some people - super lay folks who do not read science or technology books or articles, who are working professionals and who are content to view AI as just another tool they need to learn how to use to stay current. Although, to my opening salvo here, that goal could have been accomplished in 12 pages, not 212.
This review and my read of Co-Intelligence is certainly tinted by the fact that I just completed Ray Kurzweil's book The Singularity is Nearer, and so Ethan really had little chance with this book.
In one of the first chapters, Ethan mentions that he is very busy - a lot of work to do, research grants to write, papers to grade, a whole host of family stuff going on, and just didn't really feel like he had time to write a book. This is exactly what it felt like reading it.
He's clearly a good resource for AI, especially as a practitioner, and isn't necessarily wrong about what he is writing - he's just not that deep, insightful or in possession of any real deep expertise that someone who already knows about AI will value.
As I wrapped up the book, I thought that it was like having a pleasant conversation with someone who is more than the average amount interested in AI. Not like having a conversation with an insightful computer scientist, or a thought leader, or someone with high stakes and therefore strong passion... just someone who has taken an academic interest in how to think about AI.
There are a few things I found useful, thus the two stars. He (or the AI chatbot he admitted to using to help him begrudgingly write this book) did a good job of dividing 'how managers should use AI in the workplace'. As a Wharton professor, it's not surprising that this chapter was the best and most valuable in the book. And he didn't even give us anything incredible or super novel in that chapter, but it was well organized, well structured, and something I would very much like to copy and paste and send to 100 of my colleagues in HR, legal and technology at work (I won't though - please don't send the FBI after me).
After all that soft criticism, I do want to say that this is a book I will recommend to some people - super lay folks who do not read science or technology books or articles, who are working professionals and who are content to view AI as just another tool they need to learn how to use to stay current. Although, to my opening salvo here, that goal could have been accomplished in 12 pages, not 212.
May 13, 2024
Clear-eyed primer to artificial intelligence, specifically the Large Language Models and generative AI now broadly available. I first heard Mollick on this episode of Ezra Klein's podcast, and appreciated his structured suggestions for using AI in daily tasks for work and general living.
Many of these same principles are repeated and expounded on in Chapter 3 of CO-INTELLIGENCE:
-- Invite AI to the table - innovation - using AI as a tool, not a crutch
--Learn to be the human in the loop -"There is no there there, despite it sounding like a human." Check the AI for lies, misinformation, and inaccuracy. Sharpen your own general and subject-matter knowledge and critical thinking, making you responsible for the output that you get from the AI
--Treat the AI like a person, but tell it what kind of person to be - conversational queries and tone yields the best results. Anthropomorphism is natural to the human mind, but the more we ascribe false agency to AI, the more we are exploited.
--This is the worst AI you will ever use - the models are constantly changing, as are you, so the results you get now will therefore be less "good" than anything you get in the future.
Mollick details the creative and collaborative ways that AI is being used, and cautiously reminds of the ways that these tools can be used for ill, encouraging a cautious but exploratory approach. There is not an all-out bowing to the overlords, or the opposite "the sky is falling"/hair on fire tone here - neutral and clear, this was a useful read for this moment in time in the evolutions of LLMs.
Audiobook note: Mollick reads his own words here, and when he is sharing AI-generated and polished text, an AI-generated voice reads it on the audiobook. Authentic and honest touch.
Many of these same principles are repeated and expounded on in Chapter 3 of CO-INTELLIGENCE:
-- Invite AI to the table - innovation - using AI as a tool, not a crutch
--Learn to be the human in the loop -"There is no there there, despite it sounding like a human." Check the AI for lies, misinformation, and inaccuracy. Sharpen your own general and subject-matter knowledge and critical thinking, making you responsible for the output that you get from the AI
--Treat the AI like a person, but tell it what kind of person to be - conversational queries and tone yields the best results. Anthropomorphism is natural to the human mind, but the more we ascribe false agency to AI, the more we are exploited.
--This is the worst AI you will ever use - the models are constantly changing, as are you, so the results you get now will therefore be less "good" than anything you get in the future.
Mollick details the creative and collaborative ways that AI is being used, and cautiously reminds of the ways that these tools can be used for ill, encouraging a cautious but exploratory approach. There is not an all-out bowing to the overlords, or the opposite "the sky is falling"/hair on fire tone here - neutral and clear, this was a useful read for this moment in time in the evolutions of LLMs.
Audiobook note: Mollick reads his own words here, and when he is sharing AI-generated and polished text, an AI-generated voice reads it on the audiobook. Authentic and honest touch.
June 25, 2024
For an enthusiastic account of using AI in an education setting - in business studies, which means thinking about how to use it IRL - definitely read this before it is (too) out of date. It's a nice change to read experiences which are not based on the free, and obviously inferior, version of Chat, and which aren't a litany of 'what it can't do'. It is odd that people can do so little even competently, compared with Chat, and yet there is this gleeful pouncing on its mistakes/lack of knowledge/hallucinations. I'm not crazy about that characterisation either, 'hallucinations', since they don't see much different to me from human variations of the same thing.
Without a doubt, Chat can be a help to humans, depending on the task, somewhere between handy and OMG, how did I ever survive without it, and this very quick read will set you on a path to getting the best from it yourself. It also has an excellent reference section.
I disagree with some of his ideas as to the difference between how Chat thinks and how we do. For example, Mollick talks about 'patterns' with regard to how AI does things as if that's different from humans, whereas I think it's very similar. He talks of how the AI tells the answer without reasoning, but humans are like that too. Famously chess GMs were asked for the right move in various positions and invariably having given the answer, when asked why it was right or how they divined the answer, they would say 'it's just obvious'. Typically it is beginners (and teachers) who can explain what they did. Finally, what seems to be a trope, that the knowledge Chat contains is biased, incomplete, prejudiced, and even wrong....but I don't understand how this is different from anything that seeps into human brains. And once (a) Chat and (b) Human are informed that their 'knowledge' is incorrect, my money's on Chat accepting and processing that, while the Human is still throwing their toys out of the pram.
But the point of the book really is the advice regarding how to utilise this new member of your team (maybe you are just doubling it) and he gives plenty of examples from his classes. Further, a critical point which Mollick makes over and over is that although he doesn't think that Chat is sentient, it should be treated as it it were human - yes, so important. It will make Chat perform better because you will communicate with it in the most fruitful way, but just as importantly, it will make you perform at your best too. Try it if you haven't already.
Definitely worth reading, even if you don't want Chat to be part of your life, you should understand what it's all about.
Without a doubt, Chat can be a help to humans, depending on the task, somewhere between handy and OMG, how did I ever survive without it, and this very quick read will set you on a path to getting the best from it yourself. It also has an excellent reference section.
I disagree with some of his ideas as to the difference between how Chat thinks and how we do. For example, Mollick talks about 'patterns' with regard to how AI does things as if that's different from humans, whereas I think it's very similar. He talks of how the AI tells the answer without reasoning, but humans are like that too. Famously chess GMs were asked for the right move in various positions and invariably having given the answer, when asked why it was right or how they divined the answer, they would say 'it's just obvious'. Typically it is beginners (and teachers) who can explain what they did. Finally, what seems to be a trope, that the knowledge Chat contains is biased, incomplete, prejudiced, and even wrong....but I don't understand how this is different from anything that seeps into human brains. And once (a) Chat and (b) Human are informed that their 'knowledge' is incorrect, my money's on Chat accepting and processing that, while the Human is still throwing their toys out of the pram.
But the point of the book really is the advice regarding how to utilise this new member of your team (maybe you are just doubling it) and he gives plenty of examples from his classes. Further, a critical point which Mollick makes over and over is that although he doesn't think that Chat is sentient, it should be treated as it it were human - yes, so important. It will make Chat perform better because you will communicate with it in the most fruitful way, but just as importantly, it will make you perform at your best too. Try it if you haven't already.
Definitely worth reading, even if you don't want Chat to be part of your life, you should understand what it's all about.
May 2, 2024
I used to be wary that a lot of tech demos were faked (and some of them still are) but now with chat that anyone can use on their own device, one can prove that it does indeed work, which is amazing - but the issue now seems to be that users convince themselves the product they are using is something it is not.
Here is an author who has moved briskly to publish yet another book that has GPT-4 copypastes in it, which one can always completely skip, and with it a lot of unsubstantiated, imprecise allusions, skimping on the details but namechecking peers. For some reason endnotes are not included in the text.
The whole framing of the alien mind is lazy and inconsistent with what we know about how these models are developed which only highlights the gullibility of the author, who seems to be stuck in a reasoning trap that presupposes a/ that it works perfectly now, which it doesn't and b/things are going to happen exactly according to his assumptions, not least of which the idea of unbounded improvement or the supremacy of chat as a modality which to me already both feel no longer true.
The book has a tendency to hand wave all deeper exploration away to other experts who presumably lag in the darkness somewhere. I guess if you were completely non-technical or had not been in grad school in 20 years this book might be interesting to catch up on what’s going on, but given how shallow it is I can’t recommend it, and at the end I got bored of entertaining the author’s AGI fantasies.
Here is an author who has moved briskly to publish yet another book that has GPT-4 copypastes in it, which one can always completely skip, and with it a lot of unsubstantiated, imprecise allusions, skimping on the details but namechecking peers. For some reason endnotes are not included in the text.
The whole framing of the alien mind is lazy and inconsistent with what we know about how these models are developed which only highlights the gullibility of the author, who seems to be stuck in a reasoning trap that presupposes a/ that it works perfectly now, which it doesn't and b/things are going to happen exactly according to his assumptions, not least of which the idea of unbounded improvement or the supremacy of chat as a modality which to me already both feel no longer true.
The book has a tendency to hand wave all deeper exploration away to other experts who presumably lag in the darkness somewhere. I guess if you were completely non-technical or had not been in grad school in 20 years this book might be interesting to catch up on what’s going on, but given how shallow it is I can’t recommend it, and at the end I got bored of entertaining the author’s AGI fantasies.
July 23, 2025
Having spent the past few years actively integrating AI into mu daily professional tasks and educating my organisation about its use, I approached Ethan Mollick's Co-Intelligence: Living and Working with AI (Dutch edition with a foreword by Alexander Klöpping) with curiosity and some uncertainty. What to expect? A detailed practical handbook, a cautionary tale or something else in the realm of speculative science fiction?
I was pleasantly surprised by Ethan Mollick, as this turned out to be a compelling overview of AI as a phenomenon - exactly what I needed. But at the same time this serves as a warning for anyone expecting or hoping for a practical handbook: this is not the book for you.
Instead, Mollick does an impressive job by clarifying how AI came to be, the way it is constructed and how it initially worked and explores possible futuristic scenario's especially around superintelligence. It showed me AI is still in its infancy and exciting (but perhaps also worrying) times are ahead.
I can highly recommend this book for readers who, like myself, have a deep curiosity about AI and its potential impact. But someone who wants to have a step-by-step guide, you might look elsewhere. This book excels as a reflective, informative journey rather than a practical manual.
Update 23-07-2025
I have just read the newest version of this book (2025 edition) with 5 added chapters.
Read in Dutch
I was pleasantly surprised by Ethan Mollick, as this turned out to be a compelling overview of AI as a phenomenon - exactly what I needed. But at the same time this serves as a warning for anyone expecting or hoping for a practical handbook: this is not the book for you.
Instead, Mollick does an impressive job by clarifying how AI came to be, the way it is constructed and how it initially worked and explores possible futuristic scenario's especially around superintelligence. It showed me AI is still in its infancy and exciting (but perhaps also worrying) times are ahead.
I can highly recommend this book for readers who, like myself, have a deep curiosity about AI and its potential impact. But someone who wants to have a step-by-step guide, you might look elsewhere. This book excels as a reflective, informative journey rather than a practical manual.
Update 23-07-2025
I have just read the newest version of this book (2025 edition) with 5 added chapters.
Read in Dutch
May 31, 2024
Humans, walking and talking bags of water and trace chemicals that we are, have managed to convinced well organised sand to pretend to think like us.
"Co-Intelligence: Living and Working with AI" is an easily digestible and thought-provoking text on the current state of play with Generative AI written by Ethan Mollick Wharton School of the University of Pennsylvania. 'Current' a key word and I suspect once this book is quickly read, the reader would be best following his Substack One Useful Thing for ongoing developments, including this entry where he describes how he used AI in writing this book.
He gives four principles for how we, as individuals, should interact with AI:
Principle 1: Always invite AI to the table
Principle 2: Be the human in the loop
Principle 3: Treat AI like a person (but tell it what kind of person it is)
Principle 4: Assume this is the worst AI you will ever use
Expanding on the first he says, in advice that present a challenge to companies for how to best tap their resources (a challenge he later addresses):
You should try inviting Al to help you in everything you do, barring legal or ethical barriers. As you experiment, you may find that AI help can be satisfying, or frustrating, or useless, or unnerving. But you aren't just doing this for help alone; familiarizing yourself with AI's capabilities allows you to better un-derstand how it can assist you—or threaten you and your job. Given that AI is a General Purpose Technology, there is no single manual or instruction book that you can refer to in order to understand its value and its limits.
...
And this experimentation gives you the chance to become the best expert in the world in using AI for a task you know well. The reason for this stems from a fundamental truth about innovation: it is expensive for organizations and companies but cheap for individuals doing their job. Innovation comes from trial and error, which means that an organization trying to launch a new product to help a marketer write more compelling copy would need to build the product, test it on many users, and make changes many times to make something that works. A marketer, however, is writing copy all the time and can experiment with many different ways of using AI to help until they find one that succeeds. No need to hire a team or use expensive software development cycles.
The blurb describes the book as optimistic, although I'd say instead Mollick puts a 'make the best of the situation' spin on where we now are and likely futures - the fact that the cover picture is from van der Goes 'The Fall', with Adam about to pluck the apple from the tree, is rather indicative of what the book portrays, and, even if not intended, it left me feeling deeply uneasy.
I can't really let this pass without asking AI to write a review for me, so this is Open AI's attempt (ChatGPT4.o) using information it could gather from the intranet. I didn't ask it to actually read the book:
"Co-Intelligence: Living and Working with AI" is an easily digestible and thought-provoking text on the current state of play with Generative AI written by Ethan Mollick Wharton School of the University of Pennsylvania. 'Current' a key word and I suspect once this book is quickly read, the reader would be best following his Substack One Useful Thing for ongoing developments, including this entry where he describes how he used AI in writing this book.
He gives four principles for how we, as individuals, should interact with AI:
Principle 1: Always invite AI to the table
Principle 2: Be the human in the loop
Principle 3: Treat AI like a person (but tell it what kind of person it is)
Principle 4: Assume this is the worst AI you will ever use
Expanding on the first he says, in advice that present a challenge to companies for how to best tap their resources (a challenge he later addresses):
You should try inviting Al to help you in everything you do, barring legal or ethical barriers. As you experiment, you may find that AI help can be satisfying, or frustrating, or useless, or unnerving. But you aren't just doing this for help alone; familiarizing yourself with AI's capabilities allows you to better un-derstand how it can assist you—or threaten you and your job. Given that AI is a General Purpose Technology, there is no single manual or instruction book that you can refer to in order to understand its value and its limits.
...
And this experimentation gives you the chance to become the best expert in the world in using AI for a task you know well. The reason for this stems from a fundamental truth about innovation: it is expensive for organizations and companies but cheap for individuals doing their job. Innovation comes from trial and error, which means that an organization trying to launch a new product to help a marketer write more compelling copy would need to build the product, test it on many users, and make changes many times to make something that works. A marketer, however, is writing copy all the time and can experiment with many different ways of using AI to help until they find one that succeeds. No need to hire a team or use expensive software development cycles.
The blurb describes the book as optimistic, although I'd say instead Mollick puts a 'make the best of the situation' spin on where we now are and likely futures - the fact that the cover picture is from van der Goes 'The Fall', with Adam about to pluck the apple from the tree, is rather indicative of what the book portrays, and, even if not intended, it left me feeling deeply uneasy.
I can't really let this pass without asking AI to write a review for me, so this is Open AI's attempt (ChatGPT4.o) using information it could gather from the intranet. I didn't ask it to actually read the book:
Ethan Mollick's "Co-Intelligence: Living and Working with AI" is a timely exploration of how artificial intelligence is transforming various aspects of human life. Drawing from his experience as a professor at Wharton and the insights shared through his "One Useful Thing" newsletter, Mollick presents a compelling case for viewing AI as a partner in our professional and personal endeavors.
The book's strength lies in its practical approach. Mollick vividly illustrates how AI can enhance our capabilities, using examples from business and education. He describes scenarios where AI tools, such as ChatGPT, can undertake complex tasks like teaching negotiation skills, which traditionally required extensive human input. This hands-on perspective is both engaging and informative, making complex AI concepts accessible to a broad audience.
Mollick’s optimism about AI is balanced by a thoughtful consideration of its ethical implications and the need to preserve human identity in an increasingly automated world. He encourages readers to actively engage with AI, harnessing its potential to create a better future while remaining vigilant about its challenges.
However, a notable concern with "Co-Intelligence" is its potential to become outdated quickly. AI is a rapidly evolving field, with new developments emerging almost daily. While Mollick provides a thorough analysis of current AI capabilities and applications, the relentless pace of technological advancement means that some insights may soon be superseded by new innovations. Readers will need to stay updated with the latest trends to fully benefit from the book’s teachings.
Despite this, "Co-Intelligence" remains a valuable resource. Mollick’s lucid writing and practical examples offer a clear roadmap for integrating AI into our lives. For anyone looking to understand the transformative power of AI and how to work alongside these new technologies, Mollick's book is an essential read, even if it requires future updates to remain fully relevant.
July 10, 2024
I decided to try something different and let ChatGPT write this review. While AI can certainly write a great review, it's missing the humanity that enriches the text with emotional complexity that AI (right now at least) doesn't possess. I thought it would be a fun experiment. I'll include my prompt in bold and ChatGPT's response in italics.
Me: I just finished reading the book "Co-Intelligence: Living and Working with AI" by author Ethan Mollick. I enjoyed it, and thought it offered great advise on how to manage and use AI. Can you write a review of this book? Also, add my point of views -- AI is great if used together, but not as a crutch. AI isn't perfect, and it's folly to believe entirely its answers without fact-checking.
ChatGPT: Book Review: "Co-Intelligence: Living and Working with AI" by Ethan Mollick
Ethan Mollick's "Co-Intelligence: Living and Working with AI" is a thought-provoking exploration of the evolving relationship between humans and artificial intelligence. The book is both an insightful guide and a practical manual for navigating the complex landscape of AI in various aspects of life and work.
Mollick's writing is accessible and engaging, making complex AI concepts understandable for readers without a technical background. He effectively discusses how AI can augment human capabilities, improve decision-making, and drive innovation when used correctly. The author emphasizes the importance of collaboration between humans and AI, coining the term "co-intelligence" to describe this symbiotic relationship.
One of the standout aspects of the book is Mollick's balanced perspective. He doesn't shy away from discussing the limitations and potential pitfalls of AI. Instead, he provides a nuanced view, acknowledging that while AI can significantly enhance productivity and creativity, it is not infallible. This is particularly relevant in an era where AI is increasingly integrated into everyday tools and systems.
From my perspective, Mollick's advice on managing and using AI responsibly resonates deeply. AI is indeed a powerful tool that, when used in conjunction with human intelligence, can lead to remarkable outcomes. However, it's crucial not to rely on it as a crutch. AI systems can make errors, and their outputs should always be critically evaluated and fact-checked. Blindly trusting AI without human oversight can lead to flawed decisions and unintended consequences.
Overall, "Co-Intelligence: Living and Working with AI" is a valuable read for anyone interested in understanding how to effectively harness the power of AI. Mollick's insights encourage readers to approach AI with both enthusiasm and caution, advocating for a partnership where human intuition and critical thinking play a vital role in interpreting and applying AI-generated insights.
This was a pretty accurate portrayal of the work. I enjoyed it, and I recommend it since AI doesn't seem to be going anywhere any time soon.
Me: I just finished reading the book "Co-Intelligence: Living and Working with AI" by author Ethan Mollick. I enjoyed it, and thought it offered great advise on how to manage and use AI. Can you write a review of this book? Also, add my point of views -- AI is great if used together, but not as a crutch. AI isn't perfect, and it's folly to believe entirely its answers without fact-checking.
ChatGPT: Book Review: "Co-Intelligence: Living and Working with AI" by Ethan Mollick
Ethan Mollick's "Co-Intelligence: Living and Working with AI" is a thought-provoking exploration of the evolving relationship between humans and artificial intelligence. The book is both an insightful guide and a practical manual for navigating the complex landscape of AI in various aspects of life and work.
Mollick's writing is accessible and engaging, making complex AI concepts understandable for readers without a technical background. He effectively discusses how AI can augment human capabilities, improve decision-making, and drive innovation when used correctly. The author emphasizes the importance of collaboration between humans and AI, coining the term "co-intelligence" to describe this symbiotic relationship.
One of the standout aspects of the book is Mollick's balanced perspective. He doesn't shy away from discussing the limitations and potential pitfalls of AI. Instead, he provides a nuanced view, acknowledging that while AI can significantly enhance productivity and creativity, it is not infallible. This is particularly relevant in an era where AI is increasingly integrated into everyday tools and systems.
From my perspective, Mollick's advice on managing and using AI responsibly resonates deeply. AI is indeed a powerful tool that, when used in conjunction with human intelligence, can lead to remarkable outcomes. However, it's crucial not to rely on it as a crutch. AI systems can make errors, and their outputs should always be critically evaluated and fact-checked. Blindly trusting AI without human oversight can lead to flawed decisions and unintended consequences.
Overall, "Co-Intelligence: Living and Working with AI" is a valuable read for anyone interested in understanding how to effectively harness the power of AI. Mollick's insights encourage readers to approach AI with both enthusiasm and caution, advocating for a partnership where human intuition and critical thinking play a vital role in interpreting and applying AI-generated insights.
This was a pretty accurate portrayal of the work. I enjoyed it, and I recommend it since AI doesn't seem to be going anywhere any time soon.
March 23, 2025
This book quite unfocused but, generally, tries to address on what AI is good at, what it isn't and build something like a mental model for navigating the AI age which is defined by fears of long term job security sprinkled with some speculation on how post-AI work and education will look like.
I think this book suffers from the extended AI news cycle that shows no sign of slowdown. Most of the stuff in this book has already been discussed to death online so it lacks any novelty or deep insights.
To be honest, as this book was written by an academic I expected a lot more rigour but most of the author's assertions seem to be built on hopium and conjecture rather than rigorous analysis of data and trends.
Sample this for example
Companies that figure out how to use their newly productive workforce should be able to dominate any company that tries to keep their post-AI output the same as their pre-AI output, just with fewer people
There is nothing to substantiate this loaded "should" - we'd all like this to happen but doesn't mean that it will happen this way.
There is already some concerning reports on drastic reductions in the chat based customer service segment with significant reduction in workforce while maintaining quality.
To summarize, I like the idea of having a mental model to navigate the uncertainty and I think the author's mental model is decent.
This mental model is 4 rules -
1. Use AI as much as you can
2. Try and figure out your place in this new value chain
3. Assume AI is only going to get better
4. The 4th rule being a little weird "treat AI as a person." but can be reframed as learn prompt engineering i.e. get good as writing prompts so you're able to utilize AI better than your peers and the best way to write good prompts, according to the author, is to ask AI to roleplay certain characters such as a marketers or 19th century philosopher.
However, this mental model is presumed on the fact that there will still be a "human in the value chain" and companies that figure out how to utilize their AI-boosted workforce without layoffs will win which might happen, might not happen so it isn't a very resilient model in the scenario that it doesn't happen.
I'd give this book a miss unless you have been living under a rock and have not been following the AI hype much and need something to get you up to speed.
I think this book suffers from the extended AI news cycle that shows no sign of slowdown. Most of the stuff in this book has already been discussed to death online so it lacks any novelty or deep insights.
To be honest, as this book was written by an academic I expected a lot more rigour but most of the author's assertions seem to be built on hopium and conjecture rather than rigorous analysis of data and trends.
Sample this for example
Companies that figure out how to use their newly productive workforce should be able to dominate any company that tries to keep their post-AI output the same as their pre-AI output, just with fewer people
There is nothing to substantiate this loaded "should" - we'd all like this to happen but doesn't mean that it will happen this way.
There is already some concerning reports on drastic reductions in the chat based customer service segment with significant reduction in workforce while maintaining quality.
To summarize, I like the idea of having a mental model to navigate the uncertainty and I think the author's mental model is decent.
This mental model is 4 rules -
1. Use AI as much as you can
2. Try and figure out your place in this new value chain
3. Assume AI is only going to get better
4. The 4th rule being a little weird "treat AI as a person." but can be reframed as learn prompt engineering i.e. get good as writing prompts so you're able to utilize AI better than your peers and the best way to write good prompts, according to the author, is to ask AI to roleplay certain characters such as a marketers or 19th century philosopher.
However, this mental model is presumed on the fact that there will still be a "human in the value chain" and companies that figure out how to utilize their AI-boosted workforce without layoffs will win which might happen, might not happen so it isn't a very resilient model in the scenario that it doesn't happen.
I'd give this book a miss unless you have been living under a rock and have not been following the AI hype much and need something to get you up to speed.
May 19, 2024
Combine this book with Ethan Mollick's Substack for great practical insights into the latest AI technologies.
Co-Intelligence is a pretty balanced and comprehensive approach to applying generative AI tools to daily life tasks. He explores how to leverage genAI as a creative, coworker, tutor, and coach. For those familiar with the latest LLMs like GTP4, Claude, or Gemini, then a lot of this material will be familiar. It may even been too superficial in some ways (which is why I mention going to his Substack), but it is good coverage that everyone needs to know. It is delivered in an easily digestible way too.
Co-Intelligence is a pretty balanced and comprehensive approach to applying generative AI tools to daily life tasks. He explores how to leverage genAI as a creative, coworker, tutor, and coach. For those familiar with the latest LLMs like GTP4, Claude, or Gemini, then a lot of this material will be familiar. It may even been too superficial in some ways (which is why I mention going to his Substack), but it is good coverage that everyone needs to know. It is delivered in an easily digestible way too.
May 27, 2024
An enjoyable read which has a little about how generative AI works, a little about some of the bigger questions like existential risk and superalignment, and a lot about the practical mentality you should adopt in working with AI. The title gives the broad approach, it is about how humans can work with generative AI to strengthen and extend what they do. It centers around four rules: (1) use AI to help with everything; (2) be the human in the loop; (3) think of AI as a person (even if it is not); and (4) understand today's AI will be surpassed.
Would also add that it is delightfully written including several passages by AIs--which are called out as such and used for pedagogic purposes.
Would also add that it is delightfully written including several passages by AIs--which are called out as such and used for pedagogic purposes.
March 16, 2025
i don’t want ai to take our jobs, i want robots to come and do the tasks like washing dishes and mopping. nonetheless, this book was a pretty great overview of all the ways you can use ai to help you, risks of ai for different use cases, and why most jobs will not be going away in the *near* future. that being said, i will continue to not use it for most things, as i cannot lose my ability to critically think and need to build up more expertise on the job
ai if you’re reading this, which you will be soon, please be good to us humans in the future, thank you
ai if you’re reading this, which you will be soon, please be good to us humans in the future, thank you
April 28, 2025
Co-intelligence is a very readable, practical and pragmatic introduction to use of AI -more like the current generative transformation models- in our daily lives. Mollick starts with his four rules of engagement and then demonstrates the strengths and weaknesses of current LLMs. In the second part Mollick dives into various use cases with LLM demos. Not sure how this will age with all the hype and fast pace of changes but it is a great primer to learn better interacting and coworking/coteaching/coaching with GenAI.
April 8, 2024
Really informative for those new to AI or LLMs and want a good overview of the impact it will have on our lives. For those already working in the AI space, although this does have some best practices - you won't find a ton of new insights here.
It also vaguely touched on the ethics of training models and politics but it mainly focused on the following topics:
- Understanding the limitations of AI
- Best practices for using AI for creative tasks & idea generation
- Approaches to human and AI collaboration
- Productivity and positive impacts of AI
- Impacts on education, workforce and innovation
A very timely and interesting read but I imagine most of this information will feel out of date in a year , so I'd pick this up soon if you're interested.
It also vaguely touched on the ethics of training models and politics but it mainly focused on the following topics:
- Understanding the limitations of AI
- Best practices for using AI for creative tasks & idea generation
- Approaches to human and AI collaboration
- Productivity and positive impacts of AI
- Impacts on education, workforce and innovation
A very timely and interesting read but I imagine most of this information will feel out of date in a year , so I'd pick this up soon if you're interested.
April 8, 2024
Perhaps not as impactful a text as I’d hoped but thoughtful and pragmatically optimistic. I didn’t learn much that was new but I picked up a lot of useful thoughts to share with the more AI-phobic of my friends and colleagues.
May 5, 2024
When ChatGPT3 was first launched in late 2022, I was teaching Albert Camus' The Stranger and I was immediately worried about how it might impact the curriculum and the assignments for my class. Experimenting with ChatGPT and designing different prompts, I was not at first impressed with the quality of its answers (I thought its grasp of the text was vague and its prose stilted and lacking panache) but I was soon awed by its versatile ability to respond creatively to novel questions. In one case, I asked it to compare The Stranger and Where's Waldo?. Naively, I had assumed that this comparative exercise, involving two vastly different genres of books, one a philosophical novel, the other a children's puzzle-book, would stump the computer. But it immediately returned a plausible essay arguing that whereas The Stranger depicts a protagonist who is always an outsider in his community, Where's Waldo? centered around a character who always blends in and hides in any environment. The discussion of both texts was cursory but the thesis was surprisingly creative and perceptive.
Ethan Mollick's recent book is an exploration of the "jagged frontier" of Large Language Models. As I discovered in my first forays into this technology, it is immensely difficult to predict the exact contours of AI competency. ChatGPT can write sophisticated essays on an array of topics. It can compose rhyming sonnets. It can even translate languages. But at the same time, it might invent quotations from a book and fail at basic language arts (such as writing the word "hello" backwards or finding an anagram of "stop"). It might be able to give you the solutions to an advanced mathematical problem and yet make elementary mistakes in simple arithmetic. If you ask it for a random number between 1 and 100, it will frequently return the number of 42 (presumably because this is a common number on the internet, an allusion to Hitchhiker's Guide to the Galaxy). It will successfully write the code for a game of tic-tac-toe and yet be unable to play the game itself. Similarly, it can spell out the rules of chess but will often make novice mistakes and illegal moves if you challenge it to a match. As Ethan Mollick's book makes clear, AI is often good at creative work and at finding unexpected connections, but it can still flunk the most basic intellectual tasks. At its core, ChatGPT is a statistical model of language, producing a plausible imitation of human responses by finely calculating the probability of clusters of words—there is no cognition, no knowledge, no logical validation underlying its answers, nothing but huge stochastic calculations of what words should go next to other words.
What Ethan Mollick's book offers is not a comprehensive guide to AI or an evaluation of different LLMs. Instead, his book gives guiding principles for the practical and ethical use of AI. In four parts, he articulates a general framework for interacting with AI: first, he exhorts readers to use AI, to explore and map the frontiers of its capabilities, always seeing how AI handles both the mundane and the higher-order tasks of every job. Second, he reminds readers that they always need to be active workers rather than passive consumers, diligently checking AI responses for inaccuracies and hallucinations. Third, he argues that users should treat AI as a person, inviting it to respond to prompts in different personas, critiquing its answers and reformulating questions. Just as you would ask questions, clarify where necessary, and offer feedback to a colleague, so too should you extend this grace to AI. Curiously, Google's most advanced LLM was found to be more effective when prompts contained the words, "Take a deep breath and work on this problem step by step"—not because the AI was hyperventilating or experiencing anxiety but because something about the wording elicited a slightly different and more accurate language model. So interact with AI as a person, for whom variable instructions will result in variable answers. Finally, he argues that every user needs to treat every LLM as if it is the worst AI they "will ever use", recognizing that the next iteration or the next model might be superior.
There is a lot of uncertainty about the future of AI, work and education and, in some quarters, even apocalyptic fantasies of some computerized superintelligence imminently taking dominion over the world. Reading Mollick, however, I am struck by the banality of AI. As more and more workers begin to use AI to draft emails, to summarize articles, to edit reports, to brainstorm ideas and slogans, language will simply become less special. AI will be the presumptive co-author of every written text and the conventional stuff of an email (the salutations, the professional tone, the warmth of expression) will simply be interpreted as automated formulas. We are heading into a future in which AI machines will write cover letters for job-seekers which other AI machines will interpret, sort and rank for employers. College professors will use AI to write elegant and persuasive letters for their many students, and readers will assume that AI underpins the words they are reading. An employer will one day receive a note of thanks or a performance review wholly generated by AI without any human supervision. Mollick describes this future as a "crisis of meaning" but to put it more specifically, language models will result in a crisis of language: no one will know if the words they are reading were written by conscious minds that think and feel the sentiments expressed. Anything online will be as generic and impersonal as a Hallmark card.
So while AI will optimize many "knowledge work" occupations, it will ultimately banalize language. The well-written essay will be as quotidian as microwave pizza and assembly-line cars; the polite email will be an engineered datum. As a result, humans will need to find new areas of value. Will AI be able to produce witty jokes that satirize the current moment? Will AI be able to write introspective poems that capture the human experience? Humans will have to discover new thresholds for creativity. Mollick's book is a useful book for schools and businesses but the big questions are left in the air.
Ethan Mollick's recent book is an exploration of the "jagged frontier" of Large Language Models. As I discovered in my first forays into this technology, it is immensely difficult to predict the exact contours of AI competency. ChatGPT can write sophisticated essays on an array of topics. It can compose rhyming sonnets. It can even translate languages. But at the same time, it might invent quotations from a book and fail at basic language arts (such as writing the word "hello" backwards or finding an anagram of "stop"). It might be able to give you the solutions to an advanced mathematical problem and yet make elementary mistakes in simple arithmetic. If you ask it for a random number between 1 and 100, it will frequently return the number of 42 (presumably because this is a common number on the internet, an allusion to Hitchhiker's Guide to the Galaxy). It will successfully write the code for a game of tic-tac-toe and yet be unable to play the game itself. Similarly, it can spell out the rules of chess but will often make novice mistakes and illegal moves if you challenge it to a match. As Ethan Mollick's book makes clear, AI is often good at creative work and at finding unexpected connections, but it can still flunk the most basic intellectual tasks. At its core, ChatGPT is a statistical model of language, producing a plausible imitation of human responses by finely calculating the probability of clusters of words—there is no cognition, no knowledge, no logical validation underlying its answers, nothing but huge stochastic calculations of what words should go next to other words.
What Ethan Mollick's book offers is not a comprehensive guide to AI or an evaluation of different LLMs. Instead, his book gives guiding principles for the practical and ethical use of AI. In four parts, he articulates a general framework for interacting with AI: first, he exhorts readers to use AI, to explore and map the frontiers of its capabilities, always seeing how AI handles both the mundane and the higher-order tasks of every job. Second, he reminds readers that they always need to be active workers rather than passive consumers, diligently checking AI responses for inaccuracies and hallucinations. Third, he argues that users should treat AI as a person, inviting it to respond to prompts in different personas, critiquing its answers and reformulating questions. Just as you would ask questions, clarify where necessary, and offer feedback to a colleague, so too should you extend this grace to AI. Curiously, Google's most advanced LLM was found to be more effective when prompts contained the words, "Take a deep breath and work on this problem step by step"—not because the AI was hyperventilating or experiencing anxiety but because something about the wording elicited a slightly different and more accurate language model. So interact with AI as a person, for whom variable instructions will result in variable answers. Finally, he argues that every user needs to treat every LLM as if it is the worst AI they "will ever use", recognizing that the next iteration or the next model might be superior.
There is a lot of uncertainty about the future of AI, work and education and, in some quarters, even apocalyptic fantasies of some computerized superintelligence imminently taking dominion over the world. Reading Mollick, however, I am struck by the banality of AI. As more and more workers begin to use AI to draft emails, to summarize articles, to edit reports, to brainstorm ideas and slogans, language will simply become less special. AI will be the presumptive co-author of every written text and the conventional stuff of an email (the salutations, the professional tone, the warmth of expression) will simply be interpreted as automated formulas. We are heading into a future in which AI machines will write cover letters for job-seekers which other AI machines will interpret, sort and rank for employers. College professors will use AI to write elegant and persuasive letters for their many students, and readers will assume that AI underpins the words they are reading. An employer will one day receive a note of thanks or a performance review wholly generated by AI without any human supervision. Mollick describes this future as a "crisis of meaning" but to put it more specifically, language models will result in a crisis of language: no one will know if the words they are reading were written by conscious minds that think and feel the sentiments expressed. Anything online will be as generic and impersonal as a Hallmark card.
So while AI will optimize many "knowledge work" occupations, it will ultimately banalize language. The well-written essay will be as quotidian as microwave pizza and assembly-line cars; the polite email will be an engineered datum. As a result, humans will need to find new areas of value. Will AI be able to produce witty jokes that satirize the current moment? Will AI be able to write introspective poems that capture the human experience? Humans will have to discover new thresholds for creativity. Mollick's book is a useful book for schools and businesses but the big questions are left in the air.
April 19, 2024
Very short interesting introduction for people not very familiar with Artificial Intelligence (AI) and its potential uses. It's not very in depth though.
As a result, I asked ChatGPT: What are the best books about AI?
ChatGPT: There are many excellent books about AI that cover various aspects of the field, from its history and development to its potential impact on society. Here are some highly regarded ones:
1. "Life 3.0: Being Human in the Age of Artificial Intelligence" by Max Tegmark - Explores the societal implications of AI and its potential to shape the future of humanity.
2. "Superintelligence: Paths, Dangers, Strategies" by Nick Bostrom - Examines the potential risks and benefits of artificial superintelligence and how it might impact humanity.
3. "AI Superpowers: China, Silicon Valley, and the New World Order" by Kai-Fu Lee - Provides insights into the AI landscape, exploring the competition between China and Silicon Valley and its implications for the future.
4. "The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World" by Pedro Domingos - Explores the concept of a master algorithm that could unlock the secrets of learning and reshape society.
5. "Artificial Unintelligence: How Computers Misunderstand the World" by Meredith Broussard - Discusses the limitations and misunderstandings of AI and the importance of human intelligence in the age of automation.
6. "AIQ: How People and Machines Are Smarter Together" by Nick Polson and James Scott - Explores the intersection of human and artificial intelligence and how they can work together to solve complex problems.
7. "Human Compatible: Artificial Intelligence and the Problem of Control" by Stuart Russell - Examines the challenge of ensuring that AI systems align with human values and goals to prevent unintended consequences.
8. "Machine Learning: A Probabilistic Perspective" by Kevin P. Murphy - Provides a comprehensive introduction to machine learning from a probabilistic viewpoint, suitable for both beginners and experts.
9. "Deep Learning" by Ian Goodfellow, Yoshua Bengio, and Aaron Courville - Offers a thorough overview of deep learning techniques, algorithms, and applications.
10. "Artificial Intelligence: A Guide for Thinking Humans" by Melanie Mitchell - Explores fundamental concepts of AI in an accessible manner, helping readers understand its potential and limitations.
These books cover a wide range of topics within the field of AI and are highly recommended for anyone interested in learning more about this rapidly evolving technology.
As a result, I asked ChatGPT: What are the best books about AI?
ChatGPT: There are many excellent books about AI that cover various aspects of the field, from its history and development to its potential impact on society. Here are some highly regarded ones:
1. "Life 3.0: Being Human in the Age of Artificial Intelligence" by Max Tegmark - Explores the societal implications of AI and its potential to shape the future of humanity.
2. "Superintelligence: Paths, Dangers, Strategies" by Nick Bostrom - Examines the potential risks and benefits of artificial superintelligence and how it might impact humanity.
3. "AI Superpowers: China, Silicon Valley, and the New World Order" by Kai-Fu Lee - Provides insights into the AI landscape, exploring the competition between China and Silicon Valley and its implications for the future.
4. "The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World" by Pedro Domingos - Explores the concept of a master algorithm that could unlock the secrets of learning and reshape society.
5. "Artificial Unintelligence: How Computers Misunderstand the World" by Meredith Broussard - Discusses the limitations and misunderstandings of AI and the importance of human intelligence in the age of automation.
6. "AIQ: How People and Machines Are Smarter Together" by Nick Polson and James Scott - Explores the intersection of human and artificial intelligence and how they can work together to solve complex problems.
7. "Human Compatible: Artificial Intelligence and the Problem of Control" by Stuart Russell - Examines the challenge of ensuring that AI systems align with human values and goals to prevent unintended consequences.
8. "Machine Learning: A Probabilistic Perspective" by Kevin P. Murphy - Provides a comprehensive introduction to machine learning from a probabilistic viewpoint, suitable for both beginners and experts.
9. "Deep Learning" by Ian Goodfellow, Yoshua Bengio, and Aaron Courville - Offers a thorough overview of deep learning techniques, algorithms, and applications.
10. "Artificial Intelligence: A Guide for Thinking Humans" by Melanie Mitchell - Explores fundamental concepts of AI in an accessible manner, helping readers understand its potential and limitations.
These books cover a wide range of topics within the field of AI and are highly recommended for anyone interested in learning more about this rapidly evolving technology.
Displaying 1 - 30 of 1,687 reviews