Cory Doctorow, a fellow #Canadian, writes a lot of interesting stuff. I agree with his positions on many things, but not all. For example, I'm about ten thousand percent behind his opposition to anti-circumvention laws; I was one of the thousands of Canadians who wrote to the government opposing the introduction of the law many years ago.
However, his blog on Thursday, staking out the position that opposition to "AI" (LLM) is just geeky #purity culture is somewhere between "flat-out wrong" and "disingenuous at best".
My position against #LLM #slop everywhere is both because of #ethical #concerns and practical ones. There does not exist an LLM right now that was built and trained ethically; they are all statistical plagiarism machines, and speaking as someone whose #prose and #code has been plagiarized by every single one of them, that pisses me off, royally.
That's a show-stopper for me, but even if it wasn't, the #practical concerns - that the output is #untrustworthy, that the #references can't be checked, that the #code is #insecure and #unmaintainable, that the #licensing status is unclear, that it's a #copyright violation - are *also* enough to rule out #LLMs at present.
He then presents a #strawman argument - all tech is fruit of the poisoned tree, the #transistor was invented by a racist, etc. But William Shockley is not designing or manufacturing any of the transistors / #ICs I use today.
So, @doctorow - I gotta say I disagree. And that's fine.







