Fuck AI

Work on projects regardless of whether they can be monetized or not. Stand up your own forgejo so you can also demonstrate DevOps and sysadmin skills. Try to be as flexible as possible - remember the entire saying of “a jack of all skills is a master of none, but is still better than a master of one”.

When you need two weeks and two people to do the work of dozens over the course of months the number will never recover.

That's not the reality though. There's no good evidence showing any software engineering productivity gains from AI. Maybe in the future, but not now. If that becomes the case, there's also Jevons Paradox. However, I think companies will just backfill offshore.

I'm not sure if you work in tech, but I have a front row seat from people very near to me. Some people are AI-assisted or vibe coding very discrete projects that are useful, but those are the senior people who understand the limitations of the LLMs and how to put up project- and enterprise-codebase-specific guardrails during prompting.

Junior coders are producing worthless unworking or buggy code that is worse than unproductive - it is slowing down the productive senior coders who need to review it (because the junior coders don't understand enough to know what is even wrong with it). That's if it gets reviewed. Plenty of tech are allowing AI code to be directly merged in some cases, creating quite a few bug/security time-bombs that will explode in the next few years.

That doesn't even get into the non-coding AI magical thinking around UX, design, and all the related fields - people are using it because they're afraid, to please the c-suite, even though it isn't solving any problem faster or better than a skilled human.

It's a speculative bubble but with a mirage of productivity. Most tech employees are using it similarly according to the greater-fool theory, selling hype to the next fool at an optics profit. That will end sooner or later.

There are specialities that can extend his time in the industry, allowing him to build up more of a nest egg before he makes any kind of a jump.

For example, DBAs (database administrators and specialists) will not go away any time soon. Same goes for observability (tracking user behaviour and system errors). Both of those require a sense of “correctness” and intuition that AI hasn’t gotten anywhere near to mastering.

Good luck.