This was a great read. Thank you.
Somehow this is very similar to what I felt like about six months ago. The AI fatigue is real and I'm still kind of fearing to get optimized away by the next breakthrough. It's kind of an irrational fear and I know that it might never happen, but from time to time it pops up again and makes me feel very anxious. One big part of it seems to be the high prices for "the best" models and or the hardware required to run a decent local model, which I don't want to pay, but make me fear losing track (I might be able to afford it, but paying for these companies just feels wrong to me).
However, I see people I really respect fade away into the topic completely while others are able to find the right balance. So, not using AI at all takes courage. I somehow also feel that it wastes potential of being a good tool. I personally limited my AI usage to be an "explaining machine", which helps me learning instead of forcing it to do my job. This re-ignites the passion of really working on something by myself but being able to skip the sometimes painful research part totally getting stuck on a problem in many cases.
Reading documentation is important for the learning process, but sometimes it feels very dry. So being able to "interact" with the documentation and ask questions about WHY something works as it works is very helpful in many cases and I learned a lot by using it by this.
I haven't found a final answer, but not using AI at all is probably not my final personal one.
- it is going to be so bad that vast majority of these "i dont write code anymore" guys are going to be unable to write even a c++ hello world program without an LLM
- the argument these guys come up with all the time is "I dont need to"
- Big assumption there buddy, big assumption. Lets play both cases shall we.
- Case 1: LLMs infinitely improve and nobody has to code anymore. Yea well, writing a paragraph spec isnt that hard for me bro, I already do it for every project while not using an LLM
- Case 2: LLMS go bust completely for whatever reason. We have a whole generation of mass programmers and 99% of them cant add 2 numbers in c++. Guess what? I am now one of the most sought after programmers in the entire world and part of an absolute minority
Even if is true (that LLM apocalypse starts exactly today). I'd still be very comfortable with letting SOL, Opus, Fable, and some other SOTA open sourced models to write code for me.
- maybe we hit an architectural limitation that was not previously anticipated
- maybe there just isnt enough power to run anymore data centers
- could be any of the 4654545246541365456 reasons here
Case 4: LLMs continue to improve incrementally, the people who are currently getting good at directing LLMs get pretty good at directing LLMs, the people choosing to do things the old was get or remain good at that, the job market adjusts salary expectations for each group according to how useful they are likely to become in the new normal
Case 5: any number of things that you and I haven't foreseen because the future has a habit of completely bucking expectations
And what does LLMs going bust even mean? These are programs that we run, they aren't going anywhere, even when the current AI investment bubble completely implodes. The models, and their impact on programming culture, are here to stay.