AI's real strength isn't cognitive quality but its sheer availability

Ok_Estate_6746 · reddit · 2026-11-15

The author, who has ADHD, argues that publicly available generative AI isn't great at reasoning — its true value lies in availability.

He treats LLMs as "browsers you can ask compound questions to," using them to structure thoughts that otherwise pile up. He defaults to distrusting answers, offsetting errors via calibration questions, making the model doubt its own replies, and cross-checking multiple LLMs for contradictions.
He weighs AI cognition across three reference points: versus the pre-ChatGPT vacuum (any capability is infinite progress), versus models of a few months ago (the steep consumer-visible innovation curve has largely flattened), and versus human cognition (AI can be wrong, but so can a store employee — and external human cognition is often inaccessible due to time, social norms, and unwillingness).

Core claim: AI's comparative advantage isn't thinking better than humans, but being always willing to think with you.

More from AGI Musings