Demis Hassabis and the crisis of pursuing optimization without purpose

The implicit wager of AGI, which he predicts by 2030, is that intelligence, abstracted, accelerated, and externalized, can solve any problem. Disease, poverty, climate, war. Feed it enough data, add enough layers, and the machine will converge on an answer. But what if the question itself is wrong? What if the problem isn’t scarcity of means but poverty of ends?

We are now entering what I would call the Age of Infinite Means. This is the era where the constraints on doing have all but collapsed, and the constraints on deciding what to do become existential. As I see it, to use an analogy from Hassabis the chess prodigy, the chessboard is open, the processor is primed, but nobody agrees on a strategy. Or worse, everyone agrees: maximize engagement (as social media algorithms often do by promoting outrage), dominate the market, optimize the KPI (as in the generative AI content farms churning out synthetic text to game search engines). In other words, pursue optimization in a system whose original objectives have been buried beneath quarterly incentives.

Hassabis is not alone in sensing the disquiet. But unlike many in his field, he is willing to make it explicit, even if only obliquely. His remarks function less as a formal philosophical position and more as a provocation, a cue for broader reflection, not a full diagnosis. That intelligence without wisdom is just entropy with good PR. That an AGI, given a blank ethical check, might do what corporations already do: automate mediocrity at scale.