Evidence that AI is destroying jobs for young people

Posted by duck 9/3/2025

Evidence that AI is destroying jobs for young people(www.derekthompson.org)

349 points | 337 commentspage 2

esafak 9/3/2025|

Discussed earlier in https://news.ycombinator.com/item?id=45052423

dang 9/4/2025|

Thanks! Macroexpanded:

AI adoption linked to 13% decline in jobs for young U.S. workers: study - https://news.ycombinator.com/item?id=45052423 - Aug 2025 (629 comments)

zoeey 9/4/2025||

Several teams around me stopped hiring juniors over the past couple of years. It’s not that the newcomers aren’t good, it’s just that no one has time to train them. AI showed up at just the right moment to offer a convenient excuse, and companies are happy to save on the cost of mentoring. But long term, this feels like borrowing from the future. Without someone to train, there’s no one ready to step up later.

phyzome 9/3/2025||

Skimming this, I'm not sure why it couldn't be explained by the layoffs we had a couple years ago, which were primarily at tech companies (which are indeed more exposed to LLMs) and probably hit junior devs more.

iJohnDoe 9/4/2025|

Not wrong. Also want to point out the continued layoffs at Microsoft, Intel, etc.

yowlingcat 9/4/2025||

I don't get how interest rates is given at best a cursory phrase when ZIRP regime ending is one of the biggest macro events of the past several decades. Seems like it would deserve more of a spotlight.

Jean-Papoulos 9/4/2025||

Looking at the "overall" graph they provide, it's clear that jobs as a whole have flatlined since 22, and even non-exposed jobs have had a growth rate hugely reduced. If we disregard the existence of AI, it would be easy to assume this is just the regular economic slowdown impacting younger people more than older ones. I don't see any correlation work done to isolate the source, especially as a lot of the "ai-exposed" work is also very much exposed to how much money is flowing in.

farai89 9/4/2025||

I doubt the whole narrative is true, it may just be hope that we do not need to hire because we will be able to do it with AI. But reality outside of software and tech is very different. I work in an organisation that heavily pushing AI and even with that push, a typical employee is still not utilising it fully, unprepared and expecting training from the employer. There is also a disconnect over connecting org data to these tools and between developers and cyber teams.

cyberax 9/4/2025||

My experience: we're hiring for an AI engineer position and for a frontend developer position (and yes, we posted our positions here on Hackernews).

We have a stream of cookie-cutter candidates. As if they are clones of each other, it's uncanny. They typically have a BS degree in some foreign university, then a CS Masters' in the US, experience with robotics, then several years of experience in large companies.

And they completely fold during in-person coding tasks. Like, not being able to explain the difference between DFS and BFS (depth/breadth-first search). Or being able to write a simple custom metric and train a network in Pytorch.

And a similar story for the frontend developer position.

We now literally have to add more filters to not get inundated by underqualified candidates. These filters will make it harder for beginners to even _get_ to the resume review stage.

No conclusions from me, but something's been broken in the CS jobs market for a while.

ipnon 9/4/2025|

With the size of context windows now it'd be economical to dump all the resumes into an LLM and query the pool this way.

whateveracct 9/4/2025||

Short term thinking. Yes, maybe those jobs are replaceable. But how do you train the next gen of seniors?

papascrubs 9/4/2025||

The year is 2030: The junior engineers of the Adeptus Mechanicus stand in a circle around the holy monitor, chanting prayers to the Omnissiah. Their arcane books and lore guide the machine spirit, ensuring the correct holy words of blessing are entered into the holy prompt. Petitioning the machine spirit for its most truthful, secure holy answers.

Oh wait, wrong distopian future.

jagged-chisel 9/4/2025||

Not their (the companies, managers, executives) problem. You come in knowing your job, or they move to the next name on the list.

Education is an externality.

charcircuit 9/3/2025||

This article continues to propagate the misconception that young is the same thing as inexperienced.

dataexec 9/4/2025||

We can assume the two are so strongly correlated that it's not worth it to differentiate. Even though many exceptions exist.

charcircuit 9/4/2025||

I don't think it's true at all. Considering it only takes 2 to 3 years to rack up the 10,000 hours to "master" something, young people can get very good at a lot of things. The biggest barrier in my opinion are child labor laws that get in the way of people getting experience.

iammrpayments 9/4/2025|||

If you do something 80 hours per week for 2 years it totals 8320 hours? This is far from average behavior even if you have nothing else to do except coding.

brazukadev 9/4/2025||

Yes. That is why we have many 18 years old better at programming than fresh graduates. Teenagers have a lot of free time to code.

mrheosuper 9/4/2025||

I will be extremely concerned if my teenage boy spending >10 hours everyday, in 2 years, just sitting in front of computer screen.

lelanthran 9/4/2025||||

> I don't think it's true at all. Considering it only takes 2 to 3 years to rack up the 10,000 hours to "master" something, young people can get very good at a lot of things.

They can, if they practice with feedback 8 hours a day.

Typically, young people, as a group, are not famous for practicing something 8 hours a day.

This means, for the group as a whole, it is true.

notTooFarGone 9/4/2025|||

The audacity to paint child labor laws as a "barrier for experience"...

Children can work open source and rack up experience there. This is like the most humane way in any job ever to get experience as a minor.

charcircuit 9/4/2025||

In practice they will not just do open source and people will exploit them for free work since there are significant barriers to access employers able to pay market rate. Even open source itself can potentially be exploitative due to being free work.

While open source may be okay for coding, there are other skills which may not be so easy to do from your own home. In practice they will not just do open source and people will exploit them for free work

programjames 9/4/2025||

I think you are referring to

> The strategic thinking that goes into longer-horizon tasks may be something LLMs aren’t as good at, which aligns with why entry-level workers are more affected than experienced workers.

I think the article is talking in generalities, so on average entry-level software engineers have less experience with long-horizon tasks (e.g. months-long development), though there are definitely the exceptions that prove this rule.

tangotaylor 9/4/2025|

I'd be interested to see more discussion on Section 174, how engineer salaries suddenly triggered a higher tax bill in 2022. https://qz.com/tech-layoffs-tax-code-trump-section-174-micro...

Did Section 174 changes accelerate tech job losses? It's unclear from the paper because they only look at the data excluding tech firms, not only tech firms.

I want to see cause-and-effect: the cause being tax changes, the effect being tech firm layoffs, but they don't analyze the effect.

More comments...