People belive that LLM progress will become foundation of the future economy expansion the same way as microelectronics did. But for now there are few signs of that economic benefit from AI/LLM stuff. If one do math of what productivity increase tech should give in order to have positive ROI, one would be surprised how reality is far from feasibility of investments https://www.bruegel.org/working-paper/tension-between-explod.... Yes, anecdotally people tell stories how they can code twice/trice/ten times faster, or how they atumated their whole workflow or replaced support with LLM. But that's far not enough for AI investment feasibility in existing businesses (AI startups will flurish for a while on venture money). Also anecdotally there are many failed attempts to replace people with LLMs (like mcdonalds ordering which resulted in crazy orders).
So what we have is a hype on top of beliefs in progress as continious phenomena. But progress itself has slowed greately. Where are all breakthroughs which change our way of living? Pocket computers and consumer electronics (which is not a discovery rather an optimisation) and internet (also more about scaling than inventing) were the last. 3d printing, cancer treatment, robotics thought to be the new factors. Till AI/LLM. Now AIs/LLMs are the last resourt for believers in progress and technooptimists like Musk.
alt Hacker News
Adjustment: not progress, the hype.
People belive that LLM progress will become foundation of the future economy expansion the same way as microelectronics did. But for now there are few signs of that economic benefit from AI/LLM stuff. If one do math of what productivity increase tech should give in order to have positive ROI, one would be surprised how reality is far from feasibility of investments https://www.bruegel.org/working-paper/tension-between-explod.... Yes, anecdotally people tell stories how they can code twice/trice/ten times faster, or how they atumated their whole workflow or replaced support with LLM. But that's far not enough for AI investment feasibility in existing businesses (AI startups will flurish for a while on venture money). Also anecdotally there are many failed attempts to replace people with LLMs (like mcdonalds ordering which resulted in crazy orders).
So what we have is a hype on top of beliefs in progress as continious phenomena. But progress itself has slowed greately. Where are all breakthroughs which change our way of living? Pocket computers and consumer electronics (which is not a discovery rather an optimisation) and internet (also more about scaling than inventing) were the last. 3d printing, cancer treatment, robotics thought to be the new factors. Till AI/LLM. Now AIs/LLMs are the last resourt for believers in progress and technooptimists like Musk.