In my opinion, it will be at least quite a while before the spectre of mass unemployment as a result of the wide deployment of Artificial Intelligence (AI) becomes a relevant issue for Africa. In Africa we are already dealing with the problem of mass unemployment as a result of too little productivity, whereas the problem of mass unemployment from AI deployment is a problem of too much productivity if indeed too much productivity can be a problem. Even in the advanced West, the supposedly coming AI-induced unemployment is at least a few decades out, assuming that it even comes at all. However, this is a topic of increasingly global interest, that it seems worthwhile to discuss it.

There has been plenty of debate on the topic with the debaters falling broadly into two camps. The first are the Cassandras/Doomsday prophets predicting the coming mass unemployment as described in the previous paragraph. The second are those that are skeptical of such a dystopian outcome with some of them going as far as to accuse the Cassandras of scare-mongering and mere hype.

In my reading of the debate, I have noticed that people in the first camp tend to have technical backgrounds, some being actual AI practitioners, though the best of them tend to be well-rounded thinkers. Representative of this first group is Kai-Fu Lee, who built pioneering AI systems in the 1980s while working at Apple, who then took his talents to other Silicon Valley darlings like Microsoft and Google, and who currently runs a Venture Capital firm in China called Sinovation Ventures; and Martin Ford, a computer engineering graduate and brilliant author who wrote the bestseller, Rise of the Robots: Technology and the Threat of a Jobless Future, who if anything, is even more pessimistic than Kai-Fu Lee. Of course, not all technologists share their dire predictions.

Members of the skeptical group tend to be economists. Representative of this class is Daron Acemoglu, the brilliant MIT professor who co-authored the bestseller Why Nations Fail: The Origins of Power, Prosperity, and Poverty, and Guy Standing, who authored the rather thoughtful and provocative The Corruption of Capitalism: Why Rentiers Thrive and Work does not Pay, in which, as a side note, he dismisses the whole spectre of AI or even technology in general, induced unemployment. Again, not all economists are as sanguine as Daron Acemoglu and Guy Standing.

We have surveyed the battlefield, and the principal combatants. Let’s now look at the weapons being deployed by the combatants to wage the war, in this case being the arguments that are being tossed to and fro. The skeptics go first. They claim AI is nothing more than the latest iteration of technological innovations that though may cause jobs loss in the short-term, nevertheless improve productivity to the extent that new jobs are created to replace the old that have been lost.

Not so fast, say the pessimists. While the pessimists concede that the many technological innovations that have come into being since the industrial revolution got going a few centuries ago have inevitably made life better, people like Kai-Fu Lee point out that there have been many less instances of the truly big advancements, the so called General Purpose Technologies (GPT). He mentions that there are just three, which receive broad support as being worthy to be considered GPTs. They are the steam engine, electricity and ICT. He contends that AI is shaping up to be another GPT. His counter-argument is essentially that we have much less historical data (Since it’s the GPTs that really count) for us to be so confident that this time will be no different. He believes that for a better assessment of the impact of AI, we should look at the historical record of the impact on jobs and wages of these three GPTs alone. He states that while the record of the first two clearly show far more people benefitting even though these benefits took a while to show up and a relatively small number suffered the brunt of the disruptions caused by these GPTs, the record of ICT so far, in terms of impact on jobs and wealth inequality has been far more ambiguous.

Defenders of ICT might say more time is needed and they may have a point. After all, some observers note that it took the original industrial revolution, which got its start in 1760, some 85 years before it started lifting living standards for everybody (And progressive social policies played a big role in that. It wasn’t just about technology). Then again, the first computer to be actually built, the ENIAC was built in 1945 and the beginnings of computer design can be traced to the work of Charles Babbage in 1822. MIT Professor John McCarthy originally coined the expression “Artificial Intelligence” in 1956. The first AI conference also took place that year. The IT crowd could probably come back with the fact that the technologies that enabled the Industrial Revolution to take off in 1760 were decades, if not centuries in the making...this is the kind of debate I like…ferocious…with everything on the line.

Anyway, in his book, AI Superpowers, China, Silicon Valley and the New World Order, Kai-Fu Lee makes clear that he believes AI will tilt the net job creation scales to the negative side, though he tries to proffer solutions, running the gamut from a Universal Basic Income (an idea that goes back to 1960s, with vocal supporters including Martin Luther King Jr. and Richard Nixon) to remuneration for all forms of social work that will most likely be paid for by some sort of AI tax. Martin Ford is an even bigger fan of the Universal Basic Income.

In his book, Martin Ford gives alarming example after alarming example of the progress in AI and Robotics that has enabled company after company to carry out tasks with drastically reduced employment numbers compared to what they would have been like without the advances. He goes on to say that AI looks to be on the path to becoming a utility like electricity and so any new industries will probably adopt AI from the start and will be unlikely to create many jobs, at least when compared to previous industrial eras. I must admit that the arguments in Martin Ford’s book are rather impeccable. Also, though trained as a computer engineer, he shows a solid grasp of economics. However, given the complexity of the debate and fact that we will not be able to see the end game with clarity for at least a few decades, it is entirely possible for some counter-intuitive development to occur that deflect his otherwise logical arguments. I am specifically bringing this up because Daron Acemoglu points out that something of that nature has happened before, that being the wide scale deployment of Automated Teller Machines (ATMs) in the West.

Martin Ford happens to mention ATMs in passing, and though he doesn’t discuss them in specific details, the overall tone of his book and his rather generic statement about them suggests that he believes that they have played their part in the job disruption in the banking sector. Daron Acemoglu would counter that this is incorrect. In fact, he points out that the wide-scale deployment of ATMs actually increased overall employment in the banking sector. Academic studies suggest this happened because the deployment of ATMs helped reduce the cost of banking, thus encouraging banks to open up more branches and hire people who specialized in tasks that the ATMs did not automate.

From this example, Acemoglu leads us to the general observation that often enough, sufficiently productive technologies, while automating a particular line of work, thus reducing job demand in that area, will simultaneously increase job demand in other areas. Emphasis on “sufficiently productive”. He is at pains to point out that not all productivity enhancing innovations fall into this category. Some are just productive enough to automate a task without raising the demand for labor in other tasks.

So it seems the pertinent question would be into what group does AI fall? Needless to say that that will be a very difficult question to answer as it depends on a number of complicating factors. From the inherent nature of the field itself, to the skill of its practitioners, the uses to which it is put, and also the surrounding social environment that plays a significant part in how it is deployed.

For instance, to Martin Ford’s examples of how tech start-ups are using AI to drastically reduce headcount, Daron Acemoglu might ask…”is this because of the inherent nature of AI or because of the incentives that American society gives to the venture capitalists bankrolling these start-ups that forces them to focus on the short-term, hence they look for quick and relatively safe rewards from the straightforward automation of existing tasks as opposed to focusing on longer-term, riskier research that might create more jobs in the long-term?” As an example, he cites the case of health insurance. He believes scientists and engineers working on software or hardware, health workers could use to assist patients in doing their rehabilitation therapy at home after a surgery rather than in a hospital could potentially save insurance companies lots of money, improve well-being, and create new jobs. He doesn’t say what jobs but other than scientists, engineers and health workers, I think the other obvious ones in the case of hardware would be those involved in the logistics of the warehousing, distribution, sale and repair of the devices (I wouldn’t want to give the impression that software is incapable of such benign ripple effects. A good example would be the impact of computational modeling on oil and gas exploration). He however notes that the bulk of automation efforts in the insurance industry go towards the automation of the process for the approval of insurance claims, which he states that while saving money for the insurance firm should reduce headcount. He also points out biases in the US tax code, which taxes labor at higher rate than capital. Employers have to pay payroll taxes (used to finance social security and Medicare) on labor, but not on robots. This he says, encourages companies to engage in what he dubs as “excessive automation”, where companies automate in circumstances where perhaps, it would have been wiser not to.

I mentioned in my previous post that it is not a crime for a company to deploy technology in order to cut costs. In fact, the capitalist system of competition will inevitably prevail on you to do so. But cost-cutting in some areas of the economy needs to be matched by the development of new industries and hence new jobs in others for the long term health of a society. When this is not so, it is ultimately harmful to even the companies themselves because as unemployment rises it reduces the aggregate demand for products, which puts pressure on companies’ profit margins, which forces them to engage in further cost cutting and then the economy can easily find itself in a vicious cycle. I for one think that utilizing AI in emerging industries that have potential for broad transformation but have yet do so because of unresolved pain points that prevent them from going mainstream, might be a use that could to lead to creation of new jobs in adequate numbers. An example in my opinion would be renewable energy.

Personally, I can’t make up my mind on what side of the fence matters will eventually fall but, I hope this post would have helped us become slightly more informed about the issues and to also help us see that the debate cannot be easily cast into black and white. In fact, it seems to have more than 50 shades of grey…unfortunately without the hanky panky.

BEFORE YOU GO: Please share this with as many people as possible. Also check out my book, Why Africa is not rich like America and Europe

Bibliography
1. Lee, Kai-Fu. 2018 AI Superpowers, China, Silicon Valley and the New World Order. New York: Houghton Mifflin Harcourt
2. Ford, Martin. 2015 Rise of the Robots: Technology and the Threat of a Jobless Future. New York: Basic Books
3. Acemoglu Daron, Restrepo Pascual. Jan 2018 ‘Artificial Intelligence, Automation and Work’ NBER Working Paper Series
4. Standing Guy. 2017 Corruption in Capitalism: Why Rentiers Thrive and Work Does Not Pay. Hull: Biteback Publishing
5. Banerjee Abhijit, Duflo Esther. 2019 Good Economics for Hard Times: Better Answers to Our Biggest Problems. London: Allen Lane