But Anthropic would be profitable if they stop training cries
No seriously, I recently heart someone say that and it may even be correct. However I assume this only to be correct if a stop in training doesn't change the current revenue environment - which it absolutely would. I am still convinced a lot of demand for tokens is artificial in the sense that people buy tokens because thats what you do right now (and its convenient) not because the roi is so great.
I love the idea! It's so good to see people advancing the european tech landscape. Just out of curiosity: Are there more reasons for building Uruky than Kagi being located in the US and closed source ?
But are we really still learning ? I feel like we already converge to a set of use cases. Also I am always wondering if LLMs are what they promise to be, why is it so difficult to find sources of real (measurable) value ? Wouldn't a disillusion of those overpromises trigger a reduction in demand ?
Okay lets suppose all those companies are profitable if training would stop today. What if token demand is shrinking ? I think big parts of the current demand is artificially build by e.g. FOMO and marketing without real value generated by them. There is no indication in economic data about some productivity boom resulting from AI usage. Next thing is Energy costs - that will soon eat into profitability too. I don't see how this bubble can't burst.
Problem is LLMs still produce very often code that disappoints. This ereases trust in it for me and makes me always regret the decision to try it yet again. In order to cope with that I try to read and understand the code to verify. But then there is no "scaling of value produced" because its not really faster than doing it all by myself - is this a me problem ? LLMs are no doubt great for searching, generating ideas etc. but programming...?
Oh yes I see that. I got my first ThinkPad last year and the only one that wasn't loaded with "AI" nonsense was a P1. Despite the enormous size and weight it's a similar experience than old ones gave me when I tried them. I guess thats the option for me as long as possible. At some point we have to switch to something newer anyway...
The title is a bit misleading because the article says the shortage is about to increase dramatically again in the future, due to a lack of MINT skills in younger generations and an aging population. Also imo there is more than enough to do in this sector in Germany. Might therefore be rather the economic environment than some fundamental reversal.