Hacker News new | past | comments | ask | show | jobs | submit login

No general model is the frontier.

Thousands of small, specific models are infinitely more efficient than a general one.

The more narrowed the task - the better algorithms work.

That's obvious.

Why are general models pushed so hard by its creators?

Their enormous valuations are based on total control over user experience.

This total control is justified by computational requirements.

Users can't run general models locally.

Giant data centers for billions are the moat for Model creators and corporations behind.




It's neither obvious nor true, generalist models outperform specialized ones all the time (so frequently that it even has its own name - the bitter lesson)


So you think that one general model can outperform thousands of specific ones in their specific areas?


Certain desirable capabilities are available only in bigger models because it takes a certain size for some behaviours emerge.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: