Description du poste
RE: https://mastodon.social/@glyph/117605030684077976
Great thread. The problem™ with AI use is therefore not so much cognitive offloading/tool use, but the what/who/how of the offloading. I think you could add the "why" here as well: In a work context for example it might be mandated to use LLMs. Yelling at people holding onto a job - with families, pets and loved ones attached - is just cruel. Follow #GetFediHired for a while to see the desperation, before throwing everyone under the bus. (I don t know what to make of small entrepeneurs that do AI-assisted coding to create an app, for example. I guess they also have families?)
In general, there is therefore skill and restraint involved in using "AI". It s not the skill mentioned in the "skill issue" comments AI boosters make. But the skill is to use LLMs well without liquifying your brain while doing so.
While I was still using LLMs in my private life, I found that a relatively good way to handle the liquifying problem was to be the one who produces the output: Code, text. That s not a guarantee but at least you need to test your understanding.