How AI Is Changing the Job Market (What to Do About It)

I've been avoiding writing this post for months because I didn't want to be another person yelling into the void about AI and jobs. There's so much noise on this topic and most of it is either "AI will take all your jobs and we're all doomed" or "stop panicking, new jobs will emerge like they always have." Both of those framings are lazy and they're not actually helping anyone figure out what to do.

So here's my honest read on what I'm actually seeing, where I think the real risk is, and what I think you should do about it.

What's actually happening right now

The jobs that are getting hit first are the ones that were already somewhat automated in feel, repetitive knowledge work where the task is well-defined and the output is predictable. Think entry-level data analysis, basic copywriting and content production, simple software QA testing, certain customer service tiers, basic paralegal research tasks.

I'm not speculating about this. Companies I've talked to are doing it. Not dramatically laying people off and announcing it, but quietly not backfilling roles when people leave, or handling increased volume with the same headcount. The displacement is gradual and it doesn't make headlines the way a mass layoff does, but it's real.

The roles that are expanding are the ones that require directing, evaluating, and improving AI output. Prompt engineers, AI trainers, people who audit AI-generated content for accuracy and quality, people who can identify where AI is failing and fix the underlying system. These are genuinely new roles that didn't exist five years ago.

Who's most at risk

I could be wrong about this, but my read is: the people most at risk are those who are very good at executing well-defined tasks but haven't developed strong judgment about what the right task is in the first place.

That sounds abstract so let me make it concrete. If your job is "write product descriptions based on this spec document," AI can do that. But if your job involves understanding what the customer actually needs to hear, figuring out which angle will convert, knowing when the spec document is wrong, that's judgment, and AI doesn't have it reliably yet.

Early-career workers in certain fields are in a genuinely tough spot because the entry-level work that used to teach you how to develop that judgment is getting automated away. The path from junior to senior used to run through doing a lot of repetitive foundational work until you understood it well enough to move up. That path is getting shorter and narrower. That's a real problem and I don't think anyone has a great answer for it yet.

The fields I'm watching closely

Legal: the research and document review work that junior associates used to do is increasingly AI-assisted or AI-handled. Partners still need to develop strategy and relationships. The people in the middle are in the most uncertain position.

Journalism and media: I think the local and trade publication space is in serious trouble. These outlets were already financially stressed and AI makes it even harder to justify paying for commodity news summarization. Investigative reporting, beat reporting with deep community relationships, analysis pieces, those still need humans. The question is whether the business models can survive long enough to keep them funded.

Software engineering: this one gets debated constantly and honestly I think the answer depends heavily on level. Junior developers doing relatively routine implementation work are feeling real pressure. Senior developers who are designing systems, making architecture decisions, and reviewing AI-generated code are doing fine and in some cases are more productive than ever. The gap between junior and senior in terms of job security has widened significantly.

What I think you should actually do

First: learn to use these tools. I know that sounds obvious and maybe annoying to hear. But I genuinely don't understand people who work in knowledge jobs and haven't spent serious time with Claude or similar tools. It's like refusing to use the internet in 2001. You can do it, but it's a choice with consequences.

Second: get good at evaluating AI output, not just generating it. The ability to look at something AI produced and know quickly whether it's right, wrong, missing something, or good enough, that's a skill that takes practice and it's genuinely valuable right now. Most people who use AI accept the first output too uncritically.

Third: invest in the parts of your work that AI can't touch. Relationships with specific people. Deep domain expertise in a niche. Judgment that comes from years of experience in a specific context. Creativity that's tied to your actual life experience and perspective. These things compound over time and they're genuinely hard for AI to replicate.

Fourth: if you're early in your career, I'd push back on the advice to "just specialize deeper" in whatever you're currently doing. Pay attention to where judgment and strategy sit in your field and try to move toward those functions faster than you otherwise would have. The traditional slow climb through foundational work may not be available in the same way it used to be.

What I won't tell you

I'm not going to tell you everything will be fine and new jobs will magically emerge. Maybe they will, historically they have, but "historically they have" is not a guarantee and the pace of this change is faster than previous transitions. Some people are going to have a hard time.

I'm also not going to tell you we're all doomed and there's no point trying. That's just as unhelpful and it paralyzes people who should be taking action.

The honest answer is that the job market is in real flux, the impact is uneven across fields and career stages, and the people who take the changes seriously and adapt deliberately are going to be in a much better position than those who wait and see. That's not a comforting answer but I think it's the true one.