Hacker News Reader

Comment by xanderlewis | Hacker News Reader

Sorry, but comparing this to previous technology seems totally short-sighted to me (and it’s not as though you’re the first to do so). If (if) we end up with truly general AI (and at the moment we seem to be close in some ways and still very far off in others), then that will be fundamentally different from any technology that has come before.

> jobs that AI can't do.

Sure, by definition, you’ve described the set of jobs that won’t be replaced by AI. But naming a few would be a lot more useful of a comment. It’s not impossible to imagine that that set might shrink to being pretty much empty within the next ten years.

View on HN

> It’s not impossible to imagine that that set might shrink to being pretty much empty within the next ten years.

No but it’s also not impossible to imagine the opposite. AI beat humans at chess decades ago but there are more humans generating income from chess today than there were before Deep Blue.

No one pays anyone to play chess because it’s useful.

Chess players get paid because it’s entertaining for others to watch.

So your argument only shows that we can expect work as a form of entertainment to survive. Outside of YouTube, where programmers and musicians and such can make a living by streaming their work live, this is a minuscule minority.

The strongest interpretation of what you’re saying seems to be that we’ll end up in a world where everything (science, engineering, writing, design) is a sport and none of it really matters because ultimately it’s ‘just a game’. Maybe so… but is that really something to look forward to?

They get paid because the people who can't play chess professionally watch it as a mental escape from their drudgery jobs because it reminds them of their youth when they could still dream about becoming a great chess player, and then you can use marketing displayed during the chess tournament to trick them into preferring to spend the money they make from the drudgery on the adveritser's product.

Now upgrade AI to do every job better than humans so that there are no drudgery jobs. What money are they going to spend?

Not too long ago, people would come and visit the first family in the village who had installed running water, because it was a new and exciting thing to see. And yet people don't wake up every day excited to see water coming from their kitchen tap.

Think more broadly than that single example. Perhaps humans will always be interested in economic activity that involves interacting with other humans, regardless of what the robots can do.

My intuition tells me humans will always have needs that AI can't fulfill. If AI does more and more jobs, cheaper, faster, and better than humans, then the price of these services and goods are going to drop, and that means people will have more disposable income to spend on other services and goods that are more expensive because AI can't produce those (yet).

Imagine a breakthrough not only in AI but also robotics, allowing restaurants to replace the entire staff (chefs, cooks, waiters, etc) with AI-powered robots. Then I believe that higher-end restaurants will STILL be employing humans, as it will be perceived as more expensive, more sophisticated, therefore worth a premium price. What if robot cooks cook better and faster than human cooks? Then higher-end restaurants will probably have human cooks supervising robot cooks to correct their occasional errors, thereby still providing a service superior to cheaper restaurants using robot cooks only.

I agree but also think this discussion need to go deeper into its assumptions. They can't really hold in a world with AGI. Can anyone acquire/own AGI? Why? Why not? Will anyone pay anyone for anything? Will capital, material and real estate be the only things with steep price tags? What would a computer cost if all work was done by AI?

Reply on news.ycombinator.com