Even Sam Altman Thinks AGI Talk Has Gotten Silly

Sam Altman just told Alex Heath something interesting... he called AGI an "irrelevant marketing term." Coming from the guy at the center of the AGI conversation, that lands differently than when the rest of us have been saying it for years.

Here's the thing, LLMs are genuinely impressive (I use them daily, obviously). But they're still probabilistic models spitting out unpredictable outputs. Getting from that to some emergent "AGI" self... that leap is still incredibly unclear to me.

Yes, performance scales with size in interesting ways (strong superposition, if you want the technical term), and models look smarter on the surface. But we're not actually closer to replicating human understanding, reasoning, or consciousness. And we still don't have a solid definition of AGI to measure against (kind of a problem when that's the whole goalpost).

Maybe LeCun's world models get us there eventually... but probably not before we survive a third AI winter and a few business side collapses along the way (history has a funny way of repeating itself in this industry).

At this point, tossing around AGI feels like jumping the shark just to raise another billion. The industry would be better served by thoughtful product rollouts, human positive marketing, and tempered expectations... honestly, just call it Search 2.0 and let the actual value speak for itself. That, I could get behind.