icymi, local net person Cory Doctorow has decided to give up his unofficial role as the fediverse's Emotional Support Celebrity in bombastic style by coming to the defence of "open source" LLM's https://pluralistic.net/2026/02/19/now-we-are-six/ (not tagging him to avoid a pile on)

Tante has a good write up about the strangeness of his argument here
https://tante.cc/2026/02/20/acting-ethical-in-an-imperfect-world/ (again, don't want to pile on).

I also wanted to add that Cory deploy's all this rhetorical barrage in service of... having a robot proofreader. His use case is not restoring justice and dignity to the underpaid, traumatised transnational workers who built the models he uses (
https://www.noemamag.com/the-exploited-labor-behind-artificial-intelligence/), it is not helping the people of Southern Memphis escape smog and pollution, it is not protecting the mostly female victims of NCII generation (indeed, millions of private unmanaged generative machines might make that problem even harder to fight against). It is not a meaningful argument about leaving space for "good" or ethical generation - it is sophistry in service of avoiding the inconvenience of actually assessing if he's acting ethically. Tante is right to expect more from him.
Pluralistic: Six Years of Pluralistic (19 Feb 2026) – Pluralistic: Daily links from Cory Doctorow

@kalviter

'Probably half of the "errors" it points out are false positives, which is better than the false positive rate for Google Docs' grammar-checker. As someone who uses a lot of jargon, made up words, etc in his prose, I'm used to overriding my text-editor.' - aren't these systems trainable? 50% false positives seems rather high