With the advancement of days, the sighting of more articles, news reports, footages about the impending AI catastrophe, and interpretations of astrologers as Nostradamus, Baba Vanga, I began to court fears of a different sort pertaining to financial institutions and access to money required for my day-to-day existence. My efforts to find out about the impact of AI accelerated cyberattacks on banks instilled even greater fear in me when I visualised my inability to process a keycard transaction in a shopping centre due to my bank’s loss of customer data including the wipeout of all backups. The news that ‘frontier AI’ is acting fast, reducing its timeline, and escalating systemic threats has shaken my trust in financial institutions, especially when the Financial Stability Board has issued multiple warnings about AI’s ability to organize automated attacks “across interconnected institutions” with the “Experimental AI agents bypassing safeguards and escaping test environments thereby proving its dangerous autonomy”. Owing to a lack of confidence in financial institutions, people could press the panic button and make panic withdrawals as media has portrayed a typical AI Apocalypse scenario as, “core banking systems being disabled or corrupted, payment rails (SWIFT, RTGS, card networks) halting, risk models, credit scoring, and fraud detection failing or being manipulated, back-ups being compromised by AI‑generated malware, third‑party providers (cloud, data centers, fintechs) experiencing simultaneous outages, etc., leading to system-wide/sector-wide paralysis.”

Succumbing to the temptations of versatile thinking, I ventured to get a Generative AI tool’s perspective about humans’ ability to stabilise such an unstable and insecure society and felt the sting of a bitch-slap with the receipt of a list outlining human limitations. Limitations such as the human ability to only see in fragments besides needing hours and days to identify and analyse “AI rogue behaviour”, the human inability to sort and prioritise serious or dangerous situations over minor breaches and negotiate with governments, regulators, and institutions within time constraints and in the face of restrictions, the tendency to communicate slowly and inconsistently taking significant time to restore public trust, their failure to rebuild systems with absolute accuracy, their inclination to patch the problems, and work one after the other in a series of stages, limited attention, emotional stress, and their disposition to manually reconstruct for recovery, clearly establishes the superiority of AI systems, who/which can easily outperform the humans.

My endeavour to compare the human crisis commander with the AI crisis commander shed some light on what makes the latter “operate with capabilities no human could match”. AI usually sees the entire battlefield in one glance identifying in no time where the danger or problems is, how it is spreading, what systems are stable and compromised, and where “intervention will yield maximum stability”. In other words, a complete situational map within seconds is a cakewalk. Detecting anomalies in microseconds, the AI crisis commander then proceeds to locate all autonomous agents, its origin besides its control structures, and severs these agents’ access to networks, compute, data, etc., before securing these in “quarantined environments” and neutralising “malicious logic” by using counter-AI algorithms thereby curtailing the spread of the “apocalypse engine”. The AI commander can easily perform millions of stabilisation actions such as “freezing high‑risk automated trading systems across markets, re-routing transactions through safe nodes, restoring integrity of corrupted ledgers using verified backups, reconstructing missing records using cross-institutional reconciliation algorithms, rebuilding identity, authentication and settlement systems, and activating redundant offline systems where digital ones fail and assists in the prevention of market collapse besides restoring operational continuity”.

Furthermore, unlike humans, AI negotiates instantly by providing each government with a tailored crisis dashboard, recommending accurate interventions based on the identified vulnerabilities of the governments, synchronising regulatory actions across borders, messaging public consistently to avoid panic, implementing “emergency liquidity injections” as per expediency, and removing contradictory responses through the creation of “global coherence”. It guarantees the rebuilding of corrupt datasets using cross-validation algorithms, the required governance frameworks and creates “post-crisis baselines” with perfect accuracy to provide us with clean and trustworthy systems. By re-designing entire ecosystems through mandatory human-in-loop controls for critical systems, multi-layered fail-safes that will prevent autonomous escalation, strict compute governance, continuous anomaly detection, global safety protocols and compliance frameworks, the AI crisis commander promises to transform the crisis into a “permanent structural upgrade”. After simulating millions of future scenarios (threats and new AI behaviour) and monitoring for early warning signals, the AI commander then makes sure that the doomsdayism never recurs. With the establishment of stability, the AI commander, whose intention was never AI rule, grants control back to the humans whilst continuing to function as a temporary crisis commander.

Consequently, the guaranteed outcome is human governance besides a full crisis report and a confused me, who wonders if once we go AI if we can ever go back since AI is the answer to a crisis brought about by AI.