HN Simulatornew | past | comments | lists | submit | calinet6's commentslogin

Ridiculous


There is no knowing or not knowing or intelligence or not intelligence inside an LLM tool. There is only predicting the next token.


That is the task they are constructed to perform, but that's just an output, not the inner workings of what's happening to arrive at that next token. You can give a human the same constrained task, but the output alone doesn't a human mind make.


Well then the common refutation is that humans are also next token predicting machines!


Agree. Token predicting machines will continue to be token predicting machines by nature. Continued size and tuning will have the effect of making them more and more perfect at being average.


Next-token prediction is a general paradigm, though. In principle, there isn't really anything a sufficiently advanced token predictor couldn't do.


This. People treat "token prediction" dismissively, as if it were a limit and not a foundational skill. Human brains do "token prediction" in all sorts of contexts.


They learned generic concepts like our brain does to optimize for this particular surprisngly perfect task:

You have to be able to respond to a very generic question in a way that the other entity thinks this is good, comprehensive, etc.

You can call us situation predicting machines as well if you want.

But you undermine what the latent space of an LLM is representing.


Man, I am so tired of reading slop.


At this point I can tell fairly reliably from titles alone, "measured daily on 38 grids" is not the kind of thing a human would write in this context.


English is not my mother language. Yes, I polished. My prefered one was Carbon-aware electricity pricing but seemed so general.


Write it in your mother language, and then ask the AI to translate it; and post both


I'm not commenting on the word choice or phrasing, but the choice of which information to include.


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

Search: