The people who are most optimistic about AI are people who've never tried to use it for anything they're expert in. Its reputation seems to rely almost entirely on the Gell-Mann amnesia effect. I try to explain this to people; I just wish people who know something about something just ask it a question.
I keep on giving it a go our of desperation, just to get some kind of lead in something I can research manually (which I'd investigate separately).
It always disappoints me.
Here's a great example:
=> View attached media | View attached media
=> More informations about this toot | More toots from chiffchaff@tech.lgbt
@chiffchaff I worked on an AI based news engine 2 decades ago. It was obviously much more primitive tech, but I saw a similar effect and it was what ended up tanking the project.
We were using vector based AI to find and categorize obscure news items. Human editors were used to babysit and fix the results when it was wrong.
The AI outperformed the editors at first, but within a couple of months, the human editors started catching up and eventually became better than the AI. At some point, it was a waste of time to use the AI as anything more than a brute phrase matching engine to shovel stuff to the human editors.
=> More informations about this toot | More toots from sysop408@sfba.social
text/geminiThis content has been proxied by September (3851b).