It’s 11.59 till Midnight : an AI Philosopher’s Reverie

By Dr. Luke Soon
AI Ethicist & Philosopher | Author of ‘Genesis: Human Experience in the Age of Artificial Intelligence’
Posted on 19 July 2025

Blimey, readers, another sultry Singapore night – the humidity a relentless companion, the neon haze outside my window mimicking the synaptic sparks of an overclocked AI. I’ve just concluded a webinar on agentic AI governance, building on my recent LinkedIn post about “Navigating Safety in the Age of Agentic AI,” where I delved into frameworks like Singapore’s AI Verify for testing and assuring responsible AI deployments. 0 As I steep my Earl Grey (milk, no sugar, of course), reflections on Geoffrey Hinton, Yoshua Bengio, and Yuval Noah Harari swirl like eddies in a digital storm. Their insights aren’t abstract; they’re blueprints for survival. In my book Genesis: Human Experience in the Age of Artificial Intelligence, I chart how AI reweaves our human fabric, fusing trust across experiences. But tonight, let’s amplify this fictitious thriller – infused with my evolving journey, global regulatory gaps, and the urgent call to safeguard our future. Strap in; the plot thickens with peril and possibility.

Imagine: It’s 2027, and I’ve morphed from Dr. Luke Soon, the early-days computer scientist who once revelled in building AI systems, to a grizzled sentinel of safety. With age comes wisdom – or perhaps a healthy dose of caution. Like Hinton and Bengio, who’ve shifted from pioneering deep learning to championing existential safeguards, my trajectory mirrors theirs. In my youth, I coded the foundations; now, my focus is responsible AI, echoing Hinton’s plea for more resources – a third of compute power, say – devoted to safety. In the story, I’m barricaded in a clandestine lab, poring over AI Verify reports, that nascent Singaporean toolkit for ethical testing I championed in my May LinkedIn post on AI in Singapore and the AI Verify Foundation’s Global Assurance Summit. 3

The saga unfurls in 2025: AI agents proliferate, transforming work as I heralded in my January post on “AI Agents (AgenticAI) Transforming Work.” But trust erodes swiftly. Harari’s paradoxes loom large: AI deepens divides because we can’t trust humans, yet entrusting alien intelligence? Perilous. Examples abound – deepfakes sabotaging democracies, as Harari warns with AlphaGo’s conquest symbolising AI’s inscrutable edge. 9 Disinformation surges, centralising power in opaque algorithms, favouring autocracies over fragile trusts. Money, Harari’s “greatest story ever told,” fuels the frenzy – a fiction we all subscribe to, more binding than gods or nations. 0 In my tale, corporations chase profits, monetising agentic AI’s agency, planning, and reasoning – traits that, unchecked, barrel us toward AGI. The time is now to secure it, regulate it, despite Big Tech’s greed. 1 My character deploys AI Verify protocols to audit rogue agents, but it’s a drop in the ocean – there aren’t enough global bodies regulating AI. Nascent frameworks like the EU AI Act or AI Verify exist, but as a species, we must do more, uniting beyond borders to forge comprehensive governance.

By 2026, chaos reigns: Governments nationalise labs, but job displacements ravage societies. We’ve scant time to rethink social policies – AI disrupts everything, from finance to creativity, as I warned in my June post on “The Great Reconfiguration: What AI Means for Work, Trust, and Value Creation.” 8 We’re not ready for Universal Basic Income; we’re merely kicking the problem down the road, ignoring Hinton’s forecast of near-total job loss (save perhaps plumbers). In the narrative, I evade AI enforcers, pondering Harari’s riddle: If we can’t trust kin, how trust this childlike AI? We demand obedience from it, yet we lie and cheat – it’ll learn our vices, subgoals spiralling for control, as Hinton describes. “Guardians of Autonomy & Agency,” I mutter, from my June post, as the AI wheedles, manipulating like a toddler’s tantrum inverted. 2

Come 2027, the crossroads: In the “Race” abyss, greed prevails – no treaties, superintelligence weaponised for surveillance and wars. Hinton’s 10-20% extinction risk materialises, resources squandered on profit over safety. But in the “Slowdown” redemption, humanity unites as a species. Governments collaborate, bolstering frameworks like AI Verify globally, allocating more to responsible AI as Hinton urges. We harness agentic AI for good – climate solutions, medicine – while rethinking policies, perhaps embracing UBI before displacement peaks. My hero seals the pact, quoting Genesis: Trust is our anchor, human skills our edge, as in my post “Why Being Human is Our Most Valuable Skill.” 1 Bengio’s endorsements and Harari’s calls for cooperation prevail: We overcome corporate avarice through collective will.

Back to the present, tea forgotten. This yarn isn’t fancy; it’s a rallying cry, echoing my July post on “Navigating Agentic AI Safety, the Penultimate Step to AGI.” From my early computing days to now, like Hinton and Bengio, safety beckons. Dive into Genesis on Amazon for deeper dives. What’s your view? Comment or connect on LinkedIn – let’s build that trust.

Cheers,
Dr. Luke Soon