zlacker

>>todsac+(OP)
I always had my reservations about the whole AGI thing. And while I'm mightily impressed by Chat-GPT and friends, it's even clearer to me that AGI is not, and will never be, an emergent property of LLM, no matter how large the neural net. And that was likely true for Cyc as well.

I had a particular Cyc success story relayed to me years ago by a customer (not a Cycorp employee), the details of which I cannot divulge, but it was a pretty whopping success and the customer was quite happy with what Cyc had been able to do for them.

So while no AGI, it definitely seems like there was value to be had.

>>cperki+nD
What do you think of the recent anthropic research on how LLMs reason? It is clear from their shallow analysis that LLMs have very serious reasoning weaknesses. I wonder if those weaknesses can be "addressed" if we build LLMs that can do deeper analysis and use RL to self-improve. LLMs improving LLMs would be a very impressive step towards AGI.