←back to thread

Entry-level jobs down by a third since launch of ChatGPT

(www.personneltoday.com)
120 points lsharkey602 | 5 comments | 30 Jun 25 11:38 UTC | HN request time: 0.824s | source
Show context
ttul ◴[30 Jun 25 14:09 UTC] No.44423514[source]▶
I run a mature software company that is being driven for profit (we are out of the fantastic future phase and solidly in the “make money” phase). Even with all the pressure to cut costs and increase automation, the most valuable use of LLMs is to make the software developers work more effectively, producing the feature improvements that customers want so that we can ensure customers will renew and upgrade. And to the extent that we are cutting costs, we are using AI to help us write code that lets us use infrastructure more efficiently (because infrastructure is the bulk of our costs).

But this is a software company. I think out in the “real world,” there are some low hanging fruit wins where AI replaces extremely routine boilerplate jobs that never required a lot of human intelligence in the first place. But even then, I’d say that the general drift is that the humans who were doing those low-level jobs have a chance to step up into jobs requiring higher-level intelligence where humans have a chance to really shine. And companies are competing not by just getting rid of salaries, but by providing much better service by being able to afford to have more higher-tier people on the payroll. And by higher-tier, I don’t necessarily mean more expensive. It can be the same people that were doing the low-level jobs; they just now can spend their human-level intelligence doing more interesting and challenging work.

replies(4): >>44423568 #>>44423776 #>>44423805 #>>44424131 #
486sx33 ◴[30 Jun 25 14:15 UTC] No.44423568[source]▶
So basically compressing the pay scale even further …
replies(1): >>44423597 #
eru ◴[30 Jun 25 14:17 UTC] No.44423597[source]▶
Well, many people complain about pay inequality. Compressing scales is the opposite of that, so should be welcomed?
replies(2): >>44423683 #>>44423727 #
landl0rd ◴[30 Jun 25 14:22 UTC] No.44423683[source]▶
Most of those people aren’t working in highly-paid disciplines like high tech. Generally those disciplines necessarily have wider spreads. I am perfectly fine with this.

If I suddenly have to think really hard at my job all day and do terribly if I’m undersea and still get paid the same or less, I will be left pretty bitter.

replies(1): >>44423826 #
eru ◴[30 Jun 25 14:32 UTC] No.44423826[source]▶
Wouldn't AI mean you have to think less hard than before?
replies(1): >>44424232 #
1. bluefirebrand ◴[30 Jun 25 15:03 UTC] No.44424232[source]▶
No, solving problems yourself is easier than understanding solutions that AI serves to you
replies(2): >>44424699 #>>44424803 #
2. malfist ◴[30 Jun 25 15:47 UTC] No.44424699[source]▶
Yeah, but I could just not understand the AI solution and just run with whatever it give me. No effort there. If something doesn't work, I can just tell the AI to fix it.

(Not a serious suggestion, but I do see this in the wild a lot)

replies(1): >>44425825 #
3. eru ◴[30 Jun 25 15:57 UTC] No.44424803[source]▶
I'm not so sure.

An analogy: any idiot can take a calculus class today, but it took Leibniz and Newton to come up with it in the first place. (And even those geniuses didn't do it properly: it took until the likes of Karl Weierstrass and friends to put analysis on a firm footing.)

replies(1): >>44433951 #
4. bluefirebrand ◴[30 Jun 25 17:27 UTC] No.44425825[source]▶
Yes, I do unfortunately see this in the wild a lot as well

Maybe I'm the one who is ultimately a sucker, because I take too much pride in my work to do this

But I always thought that the quality of my work and my effort would be tied to my reputation, but I don't think the world works that way unless you are very well known somehow

5. sublinear ◴[01 Jul 25 13:56 UTC] No.44433951[source]▶
AI generated code in all but the most trivial cases is never production ready.

In the real world, the resulting code that correctly and efficiently solves a problem ends up being unique enough such that an AI wouldn't learn much even if you fed this result back in. It's just going to average away all the important parts that made it correct.

If real world code wasn't so sparse and time consuming to develop it wouldn't be so valuable in the first place. Not to mention the fact that the maintenance is what you really care about and where the real costs are. Application code doesn't stand still.