from Hacker News

Ask HN: AGI is here, what do you do next?

by cryptozeus on 12/25/24, 2:03 AM with 13 comments

• by proc0 on 12/25/24, 2:30 AM

  Depends on the cost. If it's very cheap I would immediately start a few startups, ideas are endless and I don't even need to have ideas. This would be proportional to how cheap it is and how many agents I get access to at once.

  Of course, realistically it would not be cheap just based on demand, let alone the resources it uses, which I think is the real reason why it won't replace as many jobs as people fear. The answer all along is going to be that humans are cheaper to employ. I suspect that this will be the case for several years as the technology matures and this should give countries extra time to adapt.

• by rvz on 12/25/24, 2:39 AM

  Given the endless announcements of so-called "AGI" being achieved, I would just prepare for the mass layoffs in 2025.

  "AGI" just means anything at this point and it is commonly associated to mean:

  "Raise more money to achieve AGI that replaces humans at almost any economically valuable task."

  That is what it really means. A Christmas present to McKinsey, Accenture and KPMG.

• by jprete on 12/25/24, 3:07 AM

  What exactly does AGI mean in this hypothetical? The term means different things to different people.

• by Terr_ on 12/25/24, 3:33 AM

  Determine if it is a person that needs legal rights and independence.

• by BobbyTables2 on 12/25/24, 5:37 AM

  Research food trucks…

• by bag_boy on 12/25/24, 3:31 AM

  Ask it which Beatle’s album it likes the best.

• by cryptozeus on 12/25/24, 2:05 AM

  My first question "What is the meaning of life" (don't give me answer as "42")

• by fragmede on 12/25/24, 3:53 AM

  Waymo is here but Uber and Lyft haven't disappeared from the cities it's live in. The future is here, just unevenly distributed. In particular, let's say they've made the model available, since it won't be open source and that's just furthering misuse if the term. Barring some miracle, most of us won't have any hardware to run it on, just like the largest releases from Meta aren't runnable on my cellphone. So it being "here" would just be a press release. But okay, let's say either I have the hardware at home, or I can rent it on AWS, what could this do, in how much time? How fast can it "read" and retain new information? What is its "context window"? How much does it hallucinate? But okay, let's assume it's roughly human-grade.

  As a regular person, I don't have an executive assistant, but as a digital agent, I'd set it loose on my email, but how do you trust a human with that access in the first place. So there would have to be some thing that gets me to trust it first. But it wouldn't be cheap to run, so how much would I want to spend on a better spam filter? $15/hr? $30/hr? Large GPU instances run far more expensive than that. Which extends to businesses. The jobs of people who work for less than GPU instances cost to run the AGI are safe. At least for now. As long as they're not being annoying.

  AGI means only human-grade intelligence, and I've met a lot of people with that. I wouldn't give a random average human a pile of my money and tell them to go wild, so why would i do that with for a digital human-grade intelligence? But maybe this hypothetical AGI is good at day trading. As long as it makes more money than it costs to run, it's an obvious thing to have it do. (Of course, with everyone else running the same bot, the remaining alpha is infinitesimal.) Alas, people that have already been doing this pre-AGI would have an edge over me and my bot, so I don't see how it could actually make money.

  Until advanced robotics come around, an AGI is stuck in the digital realm, so it can help me reprogram the misc iot crap I've been meaning to do, or various administrative life tasks for me, but you can already hire a virtual assistant from the Philippines for fairly cheap. If I wanted that I could have that already.

  So personally, I don't see my life changing all that much on day 1 of post-AGI humanity. I have an emergency fund, and while that isn't going to last forever, I'm going to be okay for a bit.

  That's a myopic view of the situation though. What about the world past me and my needs? AGI could drive cars, but that technology is already here and the sky hasn't fallen. It might, but it's going to take a while. I have friends in other industries that have been devastated in the past few years, and they've survived, though it's been rough at times for them. AGI isn't going to make everyone obsolete overnight any more than cars did for horses.

  ASI, artificial super intelligence, is where it gets interesting, because then human thinking is obsolete, just like cars made walking places obsolete. Which is to say, ASI still won't obsolete human thinking, especially while it requires insane amounts of power to run. It'll be related to things that justify the cost, and he limited by the hardware it can run on. If we look at the beginning of the computer era, there was only room for maybe 5 or 6 of those in the US? And now look where we are. So even ASI doesn't scare me on day one.

  The day they hook it up to Skynet though, it's time for me to get out of the city and away from population centers and military bases.