[jensensbutton]

The question isn't about what you'll do when you're replaced by an LLM, it's what you're doing to future proof your job. There is a difference. The risk to hedge against is the productivity boost brought by LLMs resulting in a drop in the needs for new software engineers. This will put pressure on jobs (simply don't need as many as we used to so we're cutting 15%) AND wages (more engineers looking for fewer jobs with a larger part of their utility being commoditized).

Regardless of how sharp you keep yourself you're still at subject to the macro environment.

[simianparrot]

I'm future proofing my job by ensuring I remain someone whose brain is tuned to solving complex problems, and to do that most effectively I find ways to keep being engaged in both the fundamentals of programming (as already mentioned) and the higher-level aspects: Teaching others (which in turn teaches me new things) and being in leadership roles where I can make real architectural choices in terms of what hardware to run our software on.

I'm far more worried about mental degradation due to any number of circumstances -- unlucky genetics, infections, what have you. But "future proofing" myself against some of that has the same answer: Remain curious, remain mentally ambidextrous, and don't let other people (or objects) think for me.

My brain is my greatest asset both for my job and my private life. So I do what I can to keep it in good shape, which incidentally also means replacing me with a parrot is unlikely to be a good decision.

[paulcole]

Nobody who's been replaced by a parrot thought they were going to get replaced by a parrot.

Where's your espoused intellectual curiosity and mental ambidextrousness when it comes to LLMs? It seems like your mind is pretty firmly made up that they're of no worry to you.

[simianparrot]

In another comment I briefly mention one use case I have been implementing with LLM's: https://news.ycombinator.com/item?id=42434886

I'm being vague with the details because this is one of the features in our product that's a big advantage over competitors.

To get to that point I experimented with and prototyped a lot using LLM's; I did a deep dive into how they work from the ground up. I understand the tech and its strengths. But likewise the weaknesses -- or perhaps more apt, the use cases which are more "magic show" than actually useful.

I never dismissed LLM's as useless, but I am as confident as I can possibly be that LLM's on their own cannot and will not put programmers out of jobs. They're a great tool, but they're not what many people seem to be fooled into believing that they are; they are not "intelligent" nor "agents".

Maybe it'll be a bit difficult for juniors to get entry level jobs for a short while due to a misunderstanding of the tech and all the hype, but that'll equalize pretty quickly once reality sets in.

[luckylion]

Are you though? Until the AI-augmented developer provides better code at lower cost, I'm not seeing it. Senior developers aren't paid well because they can write code very fast, it's because they can make good decision and deliver projects that not only work, but can be maintained and built upon for years to come.

I know a few people who have been primarily programming for 10 years but are not seniors. 5 of them (probably 10 or more, but let's not overdo it), with AI, cannot replace one senior developer unless you make that senior do super basic tasks.