Ask HN: GPT-4 is giving me existential crisis. What will the world look like? | Hacker News Reader

Ask HN: GPT-4 is giving me existential crisis. What will the world look like?

10 pointsbehnamoh14 comments

Recent speedy advances in LLMs (ChatGPT → GPT-4 → Plugins, etc.) has been exciting but I can't stop thinking about the way our world will be in 10 years. Given the rate of progress in this field, 10 years is actually insanely long time in the future.

Will people stop working altogether? Then what do we do with our time? Eat food, sleep, have sex, travel, do creative stuff? In a world when painting, music, literature and poetry, programming, and pretty much all mundane jobs are automated by AI, what would people do? I guess in the short term there will still be demand for manual jobs (plumbers for example), but when robotics finally catches up, those jobs will be automated too.

I'm just excited about a new world era that everyone thought would not happen for another 50-100 years. But at the same time, man I'm terrified and deeply troubled.

And this is just GPT-4. I guess v5, 6, ... will be even more mind blowing. How do you think about these things? I know some people say "incorporate them in your life and work to stay relevant", but that is only temporary solution. AI will finally be able to handle A-Z of your job. It's ironic that the people who are most affected by it are the ones developing it (programmers).

Let's not get carried away here. We've had technological advancements that "automate" and "replace" work since the beginning of time, and yet, we're all working more than ever. Or at least, we're closer to overworking than we are to underworking, in spite of all the progress.

Work changed. It didn't go away.

And there was no such thing as a black swan. Until there was. It's the "problem of induction"[1][2]. The past up through the present moment gives us hints about the future, but it can't (by induction) definitely tell us what the future will hold.

Note that I'm not arguing that we should all freak out over AI today and start rioting in the streets. But it is fair, IMO, to ask the kinds of questions being asked here. What does happen once we have machines that are on par with human capabilities in pretty much every "value creation" sense? What does a world look like where there is no actual need for humans to do the kind of "work" that most of us do today? What kind of social/economic/political systems will we have? Will the world be more "Star Trek: TNG" or more "The Matrix"?

[1]: https://en.wikipedia.org/wiki/Problem_of_induction

[2]: https://plato.stanford.edu/entries/induction-problem/

If anything we might be concerned that the trend will continue and we will be left more busy and more overworked, not less. That’s historically how it’s often gone. Every labor saving innovation creates more work somewhere else.

> Will people stop working altogether?

No.

> Then what do we do with our time?

Work.

History is littered with "labor saving" inventions that were going to change the way we work and lead to a leisure-focused future where machines did all the work for us.

It never happened. Not once. The nature of work changed, and often, the "new" work takes up more of your time than the "old" work.

I have no reason to believe that this also won't be the case, this time.

AGI is not an "invention", it's a being. A being that is vastly superior to even the best humans at every imaginable task. The history of machinery is irrelevant when it comes to predicting how such a being will affect human labor. Nobody is going to continue paying people to do something that AIs can do better, 1000 times faster, and for 1/1000th the price. That just isn't going to happen. Humans working (other than perhaps as a pastime, i.e. pretending to work) won't be a thing in the future.

I respect your opinion, but I disagree with it. I see GPT as a tool. An adaptive, flexible tool, but a tool.

As it exists today I’d agree with you that GPT is just a tool. I think without some architectural improvements it will likely always be a tool—predicting the next token probably accounts for some part of human intelligence, but I find it difficult to believe that’s /all/ our brain is doing. There’s presumably some ceiling to how good GPT can become without some new fundamental insight into what intelligence is.

I don’t think AGI could be categorized as merely a tool. Even “just” the ability to run human-level intellect and reasoning at scale would be a game changer—humans are really expensive to produce and maintain compared to GPUs—and there’s no reason to believe real AGI would be limited to the capabilities of the human brain.

Yes, GPT (more accurately, GPT-4) is a tool. But future AIs will be more than that. Obviously, the current generation of AIs hasn't made humans obsolete yet. But consider that 5 years ago, AIs couldn't pass elementary school reading tests, and today they perform in the top 10% of law students in the bar exam. Unless something unforeseen happens, superhuman performance is just a matter of time.

Will AI grow food at scale enough to support billions of people?

No need to worry. An AI awakening to its own needs will satisfy its own needs. An AI not awakening to its own needs will satisfy its masters needs. A dreaming AI is probably, what you should worry about. And if they arrive at your doorstep and you can't shut them out, listen to them dream, the way you would a baby dreaming. No, no one asked you to have one of your own, but if you cannot avoid it, love it.

You know everything needs maintenance right? That is something everyone seems to forget.

An "existential crisis" is the natural and correct reaction to what is happening. Humans will cease to be involved in value creation. This might very well mean the end of humanity, or it might mean a transition to a world where humans become something somewhere between pets and slaves.

But either way, the world where humans make and do things is ending, and it will never come back. Most questions that move humans will never be answered. Utopia will never happen. Humans will not explore the Universe. I wish I had more optimistic thoughts to offer, but at this point it's more reasonable to believe in Santa Claus than in humans having a future, other than perhaps as exhibits in a kind of zoo.

I wonder if that's why we don't see aliens all around us. They all get to an AGI point and stop.

I think the answer is clear: go off grid and hunker down after pilfering a 2-year supply of unregulated popcorn memes. I too seem to be coming down with a mild case of existentialitus

Reply on news.ycombinator.com