Hacker News new | ask | show | jobs
by maxander 3040 days ago
> These changes are coming, and we need to tell the truth and the whole truth. We need to find the jobs that AI can’t do and train people to do them. We need to reinvent education. These will be the best of times and the worst of times. If we act rationally and quickly, we can bask in what’s best rather than wallow in what’s worst.

Saying "find the jobs that AI can't do" like it's this obvious thing; we haven't the slightest clue what AI can't do in five years. Practically yesterday Go was the go-to example of what computers couldn't do, and now it's the go-to example of how they kick our ass. Well-informed predictions for widespread self-driving car deployment range from "in five years" to "half a century at best." And the jobs that aren't threatened by AI, and that are accessible to most people, are threatened by much "dumber" kinds of automation. Human-facing "empathetic" jobs are often considered safe, but what positions actually need a human, anyway? People are happy to do a bit of extra work to get "expert advice" from a database when it comes to travel, law, or finance; stores are happy to let customers use automated checkout rather than talk to a cashier. Dextrous, highly-physical work is still safe (pending advances in robotics, which seems to move a bit slower), but only until someone figures out a way to offload the dexterity-requiring bits onto a factory, or render the whole thing obsolete. And finally there's second-order effects of the resulting economic turmoil; if most of the middle class can't count on keeping their jobs, not enough people are going to be ordering fancy lattes to pay for your role as a barista. Planning a relatively low-skilled career (anything less than a four-year technical college degree) is walking a minefield.

Us highly-educated technical generalists are probably fairly safe, but not everyone wants to study for 6+ years after highschool- and regardless, they could expect to have to relearn all the applied skills in their chosen field when the tool they originally studied gets replaced by something completely different and orders of magnitude better. The old model of "learn a skill, get a job applying that skill, eventually retire" will no longer reliably work on any point of the scale.

1 comments

It's pretty easy to guess. I just did my laundry and folded my clothes. Something "ai" can't even think about doing, not even close.
Uh huh?

https://foldimate.com/ https://www.theverge.com/2018/1/10/16865506/laundroid-laundr...

(and yes these are buggy and expensive now, but 5 years?)

5 yrs is not that long of a time. Foldmate example you linked supports my case not yours.
Actually it can but the cost to build it currently is not worth the return. In 5-10 years I don't think that will be true. First such machines will be adopted by large hotels etc where laundry which for them is usually bedsheets and towels as the tech gets smaller and better adopted by motels and over time by common people. I don't think it will be very long from when laundry room in a building or laundromats/dry cleaners will be automated where you drop off your clothes and collect them back washed dried and ironed. Almost all work in the end can be automated the biggest hurdle currently is the cost building a specialised robot for a specific task is not cheaper than humans doing it specially for housework but for how long.
Your comment doesn’t seem so different to when people were saying that ai can’t even think to beat a human go player, not even close.