I’m working on an AI policy for my org that allows us to opt out of AI note taking and prohibits AI in our comms/storytelling.

[tootbrute@fedi.arkadi.one]

@arod @seachanger great list of the reasons to not use AI.

[saorsa@neondystopia.world]

My policy for AI in programming tends to be applicable in other areas as well.

• AI should be used to assist, enhance and not replace existing workflows.
• When using AI, be sure to split your workflow into smaller, more manageable chunks.
• Proofread then validate the output against other sources before implementing it into your works.

This ensures that anything contributed by an AI meets the same expectations as a human performing the same task. If you implement the following or some variant of it into your workflow, you'll find that a lot of the common pitfalls with AI can easily be avoided. While the efficacy of AI can never been guaranteed, I find that sticking to those guidelines can help direct the output into something less liable to be derivative.

@seachanger@alaskan.social

[juandesant@mathstodon.xyz]

@cafechatnoir @seachanger pinging @WeirdWriter, who put in beautiful, powerful words how that experience of “semantic ablation” affected his writer friend. At least, it seems to be recoverable, but at what cost…

[weirdwriter@caneandable.social]

@juandesant @cafechatnoir @seachanger Yay thank you for tagging! My narrative is at the end. I’ve seen it have a drastically negative psychological consequences for everybody that uses it. Writers, readers, anybody really. I recently had a scenario where a trance friend of mine just quit writing all altogether because, on the one hand everybody was praising her for doing such a fantastic job with prompting the thing when she never used an LLM at all. The truly horrifying thing was, the positive comments were more disturbing because they praised an LLM for creating it when she never touched an LLM in her life. I’m going to write about it, but right now, the emotions are swirling around and I need to calm down after these incidences AnyWho, but if you have not read it yet, the first story is https://sightlessscribbles.com/the-colonization-of-confidence/

[thankfulmachine@oldbytes.space]

@seachanger https://www.media.mit.edu/projects/your-brain-on-chatgpt/overview/

[thankfulmachine@oldbytes.space]

@seachanger https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/

[kdwarn@social.coop]

@seachanger been collecting news articles here: https://kdwarn.net/programming/links#AI%20Sucks

[girlbandgeek@mastodon.social]

@seachanger @janeishly I really like what you're doing here. You may want to add that there is little transparency around the training data. Many models are trained on data that contains harmful biases and prejudices against BIPOC, LGBT+ people, etc. Also may involve exploitation of labor in undeveloped countries to assist with training. Good luck with getting a strong policy approved

[stonsoftware@mastodon.social]

@seachanger you miss the bias in training these models

[karlheinzhaslip@climatejustice.social]

@seachanger lol, tbh, I think the way here is to have chat-GPT hallucinate sources, provide those and let the board figure out that you just gave them another reason...