Most people don't understand how LLMs work, especially investors, everyone is just extrapolating hypotheticals and saying if it can do this it must be able to do that, if we throw more compute and energy at it, but eventually the diminishing returns have to come, the LLM has its upper limits of what it can do, i honestly think we've already reached that and that more compute and improvements will only generate marignal efficiency gains
Yes these LLMs can still be applied to many things that can still remove labour costs, but that takes time.
This is just the first boom cycle for this tech and we will have a bust and then a lull and someone will find some new version or application for it and we go again. I think he is right about this bust, but when it happens is anyones guess, the AI storys ablity to raise capital without showing enough revenue growth seems to still be on
Until I see those AI niche/edge case startups start to go tits up, I think this continues
Most people don't understand how LLMs work, especially investors, everyone is just extrapolating hypotheticals and saying if it can do this it must be able to do that, if we throw more compute and energy at it, but eventually the diminishing returns have to come, the LLM has its upper limits of what it can do, i honestly think we've already reached that and that more compute and improvements will only generate marignal efficiency gains
Yes these LLMs can still be applied to many things that can still remove labour costs, but that takes time.
This is just the first boom cycle for this tech and we will have a bust and then a lull and someone will find some new version or application for it and we go again. I think he is right about this bust, but when it happens is anyones guess, the AI storys ablity to raise capital without showing enough revenue growth seems to still be on
Until I see those AI niche/edge case startups start to go tits up, I think this continues