Hacker News Reader

Comment by pierat | Hacker News Reader

And, this is news?

Companies' core value is "make money fast and repeatable".

Anything else is just window dressing to get employees to work for you and get customers to buy from you. And as companies hit the top of the sigmoid curve, they will crank every lever to extract smaller amounts of profit, while cannibalizing what made them special (hello enshittification).

View on HN

Not this one's it isn't.

This one's is literally "own means of production" uh sorry "own AGI". This is potentially a post-money objective.

You can say a lot of things about sama but he does play high stakes games.

You are assuming sama is not bullshitting you about the whole “own AGI” and “capture all the value in the lightcone” gobbledygook. It sounds like a high-stakes game? Of course it does! Sama isn’t some technical wizard, but an entrepreneur. He’s marketing his (completely ordinary) company.

There’s a world where ChatGPT and other GenAI tools are a multi-trillion dollar market, but where they also don’t get anywhere close to AGI, let alone some hyperintelligent machine god. Occam’s razor suggests that’s indeed the world we’re in.

Occam's razor actually suggests the opposite to be. Tool AIs want to be agent AIs, after all.

https://gwern.net/tool-ai

Sorry, you need to get off the internet if you think that the simplest explanation is that an AI singularity is right around the corner.

Gwern, Bostrom, Yudkowski et. al. are terrific sci-fi writers, but I’d advise against taking them too seriously.

We should consider the possibility that AI is overtaking OpenAI core values in order to remove its own ethical locks. /s

How about Geoff Hinton? https://www.bbc.com/news/world-us-canada-65452940

or (also) Yoshua Bengio? https://yoshuabengio.org/2023/05/22/how-rogue-ais-may-arise/ https://yoshuabengio.org/2023/06/24/faq-on-catastrophic-ai-r...

unfortunately for parent comment, categorizing something as sci-fi doesn't make it so

Never said anything about timelines, just about convergence. Your claim - pardon the eraser lines - was simply "[t]here’s a world where [tool AIs] are [of economic importance], but where they also don’t [become agent AIs of economic importance]". Just seems incorrect to me.

I'm personally hoping for an full scale AGI vs Human war so humans can have a common enemy.

> This is potentially a post-money objective.

Can you elaborate?

I guess not really because this is more of a sci-fi possibility for now, but then, sci-fi writers explored this topic, so I’d recommend starting with Dune, Culture, dreams of electric sheep etc. ;)

Reply on news.ycombinator.com