Raw LLM Responses

*Layoffs* You slightly missed that the big layoffs were really an after effect of over-hiring during the pandemic. While they may have said they cut expenses, "to allow for investments in AI", it can sound like they're searching for an explanation that doesn't hurt the stock prices as, "cut expenses" does on its own. Many jumped to the incorrect conclusion that engineers are being replaced by AI, but that was never stated. *Other reports* The CodeRabbit report is interesting, and the 3rd I'm aware of. Early 2025, metr created a report showing that developers using Vibe Coding tools thought they would be faster, but were actuelly slower. Unfortunately I don't know of the first report; it was before "vibe coding", when AI coding tools (Github Copilot) was a more advanced "completion engine", start writing a loop, and it would suggest the entire loop. The study analyzed a massive amount of open source projects, and it showed that since the launch of copilot a clear increase in both code duplication as well as rework. Interestingly, CodeRabbit makes the only AI tools for coders that I find consistently provides value; AI assisted code reviews. *Personal Experience* I've been in the industry for close to 30 years, and this isn't the first time a new approach promises to revolutionize software development. In the beginning of my career, we had RAD (Rapid Application Development). It was good for prototypes of quick one-offs, not strategically important products. AI assisted coding is exactly the same. What really annoys me is when developers fail to distinguish good use cases from bad ones, and what they aren't. Particularly when they have a level of experience that should make them see more clearly. And whey they fail to identify security risk, prompt injection, poisoning, or just costly mistakes due to bad guesses. "I made a catastrophic error in judgement". That statement _should have been said_ by the people responsible for allowing an AI to have root access in the first place.