@ACM_Ethics That’s harm at an individual level, but @Crell’s blog post points to harm on much larger scales. Here are only a few of those (links from his post):
* https://www.salon.com/2024/01/09/impossible-openai-admits-chatgpt-cant-exist-without-pinching-copyrighted-work/
* https://www.independent.co.uk/independentpremium/long-reads/behind-the-ai-boom-an-army-of-overseas-workers-in-digital-sweatshops-b2401142.html
* https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/
* https://www.sfexaminer.com/news/technology/ai-water-use-study-plastic-bottles/article_de4ac7e4-b5a9-4ba7-8b5d-226ba5f8be2d.html
Why do computing professionals and the organizations that represent our interests continue to ignore these harms?
All of these are in direct opposition to the @ACM_Ethics’s Code of Ethics—and, I’m sure, to other organizations’ codes of ethics. Why, then, are these organizations not sounding the alarm? Are these codes of ethics just feel-good words with no weight behind them?