Now it's LLMs. Same old.
LLM’s have similar if not worse trade offs imo.
With an LLM, you get a cycle every 2-15 minutes. And LLMs are improving faster than outsourcing companies’ employees.
LLM’s are also much better at faking it in remarkably unpredictable ways, and are hitting diminishing returns (or even backsliding) on improvements.
And LLM’s can’t actually do anything without humans structuring everything and doing stuff on it’s behalf.
Weaker engineers and junior engineers are in more the situation you describe. This is tough and I feel for these folks but it is possible for many people to become stronger engineers if they choose to put in the work.
I'd encourage you to not take on a feeling of hopelessness here.
Anecdotally - I’ve been through 2 technical screens where they asked me to use an AI prompt to solve a problem. For one, it was a pretty trivial problem (running an hmac over some values) so I just solved it directly. They asked why I didn’t use AI, and I told them honestly that I’ve done something like this hundreds of times, why would I use a prompt for it? Didn’t make it to the next round. Now it’s totally possible that I didn’t make it because of something else. And maybe those were outliers, but it seems like I’ll need to brush up on prompting…
Don’t want the job for the salary offered? Too bad. Hire a cheaper person armed to the teeth with the best LLM coding tools and move on.
Unless you’re coming in with significant clout that will move revenue and relations to bridge partnerships across other companies, you will not be worth the extra $250k on skills alone.