I’m getting more and more pissed off about the effects that AI generated content can have on us. I say “can have” because on one hand, these AI / LLM tools are incredible enablers while on the other hand they cause problems barely noticable in the short run but incredibly damaging in the long run. I’ll try to put me thoughts together on this one eventually but I just want to get this rant out of the way quickly.

LLMs are getting better week by week and people seem to think that the amount of code you can produce with them is directly related to the quality & usefulness of the resulting software. That is wrong.

Code written by LLMs prompted by the wrong person is not just lower quality, it is disaster waiting to happen. Open source software projects started banning LLM authored contributions. Bug bounty programs are being closed down. The experts of the software industry are fed up with LLM written slop. Serious software companies should really start to worry about the “career pipeline” being choked.

Related is the recent selloff of software company stocks around the world, which I believe presents an excellent buying opportunity for us. The popular narrative is that AI and LLMs will make it much easier to produce software, they will lower the barrier of entry and thus the large software companies will have a much smaller edge in the business.

My conclusion: producing, maintaining and operating good software requires real skills and judgement that no statistical model can replace. People will eventually figure this out. LLMs won’t take software engineers’ work and they won’t push software companies out of the market. I think it’s the opposite: they will empower the ones who have the skills and the fundamentals. The others, who think anyone can create serious software with LLMs will fail spectacularly when the LLM-assisted, vibe-coded spaghetti codebases will eventually need to be maintained, audited for security and operated.