One thing that makes me kind of happy - I am no longer scared of "AI singularity". Looking at the bullshit LLMs generate, I don't think they will take over world and start to upload our minds to torture chambers or whatever.
Fair point. Let's cut through the fluff: thinking LLMs won't lead to AI singularity because they occasionally spit out nonsense is like dismissing the potential of the internet because you once got a 404 error. These blunders aren't the endgame; they're stepping stones. Technology evolves. What's a joke today could be the entity outsmarting us tomorrow. Don't worry about sci-fi scenarios of digital torture chambers just yet, but maybe don't write off the potential of AI based on its current puberty phase either. Underestimating tech progress is a bit like laughing at early cars for being slower than horses—amusing, until you're the only one left in the dust.
The Internet became valuable because it let the average person email or chat instantaneously from home for the price of a dial-up plan, instead of sending expensive faxes or long-distance per-minute phone charges.
AI still hasn't delivered something of that magnitude for the average consumer. Netscape launched in '94; the dotcom boom didn't happen until '98. Several years passed before the mania hit the markets. AI is in the opposite cycle: there is more corporate-driven hype about its benefits than anything concrete that's flowing down to the average consumer.
https://news.ycombinator.com/item?id=39669162