While billions of dollars follow the AGI prophecy, recent studies show humans outperform AI at problem-solving. A case for curious questioning instead of believing grand narratives?
There is something almost religious about the way AGI is discussed. Not just as a possibility, but as an inevitability. As a prophecy to prepare for, to align with, to invest in before it's too late. The language of inevitability has a way of foreclosing questions rather than opening them.
The studies nobody's talking about
Earlier this year, several research papers quietly contradicted the prevailing narrative. In structured problem-solving tasks requiring multi-step reasoning and contextual judgment, human teams consistently outperformed the most capable AI systems. Not by a little. By significant margins on the problems that most resembled real-world complexity.
These findings didn't make headlines. They don't fit the story. The story is one of inexorable machine progress, of a curve that only goes up, of a moment approaching when the scales tip permanently.
The prophecy has a way of making its own believers incurious. If the outcome is predetermined, why examine the evidence carefully?
What curiosity looks like instead
I'm not arguing that AI isn't powerful or that the concerns about AGI are unfounded. I'm arguing that certainty — in either direction — is epistemically dangerous right now.
The honest position is something like: we don't know. We have impressive demonstrations and surprising failures. We have capabilities that seem to exceed human performance in some domains and fall dramatically short in others. We have a technology moving fast enough that last year's benchmarks are already outdated.
What curiosity looks like, in this environment, is holding the uncertainty. Asking what the evidence actually shows. Being willing to update — toward more concern or toward more equanimity — based on what we find.
That's harder than prophecy. It offers no investment thesis, no confident prediction, no story to tell at dinner parties. But it's how you actually figure out what's happening.