←back to thread

170 points bookofjoe | 1 comments | | HN request time: 0.244s | source
Show context
lenerdenator ◴[] No.43644555[source]
> One wonders what Asimov would make of the world of 2025, and whether he’d still see artificial and natural intelligence as complementary, rather than in competition.

I mean, I just got done watching a presentation at Google Next where the presenter talked to an AI agent and set up a landscaping appointment with price match and a person could intervene to approve the price match.

It's cool, sure, but understand, that agent would absolutely have been a person on a phone five years ago, and if you replace them with agentic AI, that doesn't mean that person has gone away or is now free to write poetry. It means they're out of an income and benefits. And that's before you consider the effects on the pool of talent you're drawing from when you're looking for someone to intervene on behalf of these agentic AIs, like that supervisor did when they approved the price match. If you don't have the entry-level person, you don't have them five years later when you want to promote someone to manage.

replies(6): >>43644608 #>>43644648 #>>43644670 #>>43644681 #>>43644779 #>>43645003 #
baxtr ◴[] No.43644648[source]
The 1980 version of your comment:

>Just saw a demo of a new word processor system that lets a manager dictate straight into the machine, and it prints the memo without a secretary ever touching it. Slick stuff. But five years ago, that memo would’ve gone through a typist. Replace her with a machine, and she’s not suddenly editing novels from home. She’s unemployed, losing her paycheck and benefits.

And when that system malfunctions, who’s left who actually knows how to fix it or manage the workflow? You can’t promote experience that never existed. Strip out the entry-level roles, and you cut off the path to leadership.

replies(3): >>43644735 #>>43644743 #>>43644783 #
1. Philpax ◴[] No.43644783[source]
Not quite comparable; these systems will continue to grow in capacity until there is nothing for your average human to be able to reskill to. Not only that, they will truly be beyond our comprehension (arguably, they already are: our interpretability work is far from where it would need to be to safely build towards a superintelligence, and yet...)