That's a fair point. For people already finding great utility in GenAI it seems like a foregone conclusion, but for others it can be understandably frustrating that people take it as an article of faith that "AI is worth it."
Disclosure, I'm very much in the former camp, but I try to ground myself with broader empirical evidence. I've found an increasing amount of empirical evidence that GenAI is providing value more than commensurate with its costs.
Largely the studies I've looked at focus on productivity boosts, I guess because that is very easy to tie to economic impact. This recent thread has some relevant sources and extremely rough numbers, but the outcome seems to be that for a 1% increase in datacenter usage we may have gotten a 1.2% boost in national productivity: https://news.ycombinator.com/item?id=45794907
Now, this absolutely not a rigorous analysis, but could be a good starting point for how to think about these tradeoffs.
Disclosure, I'm very much in the former camp, but I try to ground myself with broader empirical evidence. I've found an increasing amount of empirical evidence that GenAI is providing value more than commensurate with its costs.
Largely the studies I've looked at focus on productivity boosts, I guess because that is very easy to tie to economic impact. This recent thread has some relevant sources and extremely rough numbers, but the outcome seems to be that for a 1% increase in datacenter usage we may have gotten a 1.2% boost in national productivity: https://news.ycombinator.com/item?id=45794907
Now, this absolutely not a rigorous analysis, but could be a good starting point for how to think about these tradeoffs.