Post #3835472
2026-07-15 15:25 UTC
Machine learning has been around for several decades. The current versions are presented in a particular SAAS model. We are asked to use these dubious corrupt dice-factories like the Greeks used the Delphic oracle.
But that is only one implementation of the tech. You could build a home model based only on your own work, and selected out-of-copyright sources(Shakespeare, US gov publications, etc). But that wouldn’t be as *convenient* as paying Musk/Grok, Etc $40+ bucks a month to spew slop. 1/n
Replies (1)
-
@adamvs1@mastodon.world 2026-07-15 15:25
Doctorow points out that when the current miscreants go belly up, the technology will be seen for what it is: a handy heuristic reference based on sources that you provide yourself. No need to sort through peta-bytes of data in a server farm. It can run locally and do the job. It can even improve as you give it back your edits of your own work. But that means thinking, and not racing to the false efficiency of letting a complex gambling machine replace your judgement with electronic dice.2/n