The heads of my department at work have been pushing AI coding tools, but the internal adoption rate hasn't been as strong as they hoped. They've picked someone to go around and do interviews with various engineers to try to get an understanding of the various internal opinions on these tools. I'm being interviewed tomorrow, and want to impart my interviewer with the strongest negative takeaway on the subject I can muster. The more points and sources for those points I have going into this, the better.

So fedi: what am I missing here? Anything noteworthy that I should include in my list?

- The power and water consumption of AI products should be reason enough to not use them.
- The fact that they're built with stolen materials should be reason enough to not use them.
- My main value add at the company is not the speed at which I write code, but my ability to understand the team's code.
- Anthropic released a paper recently showing that using AI coding tools harms understanding.
- There was that paper last spring that showed that programmers consistently overestimate the productivity impact of AI coding tools, and the impact is actually negative.
- Key copyright issues around AI generated code remain unresolved, which is a massive legal liability.
- The entire industry is sitting atop a massive bubble, it will be extremely painful for us when it pops, and us using these tools adds more fuel to the fire.
- These tools are heavily subsidized right now, and there's no way that lasts forever. Usage fees will surely be massively increased in the future.
- If these truly were wonder tools of massive productivity boosts, we'd see the gains in the open source community, but to date this hasn't happened. There's not a massive boom of new and successful open source projects.
- The wave of slop issues and PRs filed against open source projects demonstrates how these add little value and a ton of noise.
- The wave of scrapers making the lives of website owners hard should be reason enough to not use these tools.
- People internally have already openly complained about low quality AI code from their colleagues, we have no reason to believe the issue won't get worse if adoption increases.
- We have no evidence to believe the narrative that these tools will improve radically over the coming months/years.
- It's bizarre to have a top-down push to use a specific tool. It's always been up to the engineers to use what tools they feel make them most productive. If I think I'm more productive without AI coding tools, and my work performance is still good, then just leave me alone. No one has been trying to pressure me into using the Rust language server, for example.
- The way that AI coding tools are consistently used to justify massive layoffs across the industry leaves a really bad impression that the point of all of this isn't productivity, but to punish workers. My team has been really understaffed for a couple years now because of layoffs, and this is not helping.
- Frankly all of the ethical and practical reasons to avoid these tools are really clear to me, and the way many of my coworkers have been uncritically adopting these tools is both really depressing and makes me think less of them.