Tacitus muses on the algorithms that Google uses for suggesting new Youtube videos. Unlike all the breathless stories about how "the AI is alive", he's not impressed. He thinks he does a fair job of intentionally confusing the algorithms:
Back in the early days of the internet I used to salt my profiles with various fake information just to see if I would get ads in Portuguese, spam email from tailors in Denmark, stuff related to Icelandic scrimshaw art. The only entry on the above list that might still be carrying over would be the Bossa Nova music! I of course have all "notifications" turned to the OFF setting but another account in the household did not. Maybe that's where gardening came from.
But it seems to not matter. My efforts to punk the algorithms seem unnecessary, they autopunk themselves.
I've noticed that myself in online ads, and even on Facebook. The ads I get now are less relevant than they used to be.
ReplyDeleteForget the hype, what we have now is nowhere near true AI, regardless of how many people want it to be.
Ask an AI system "hey, wazzup". Then ask "what's the difference between a duck". Then ask "howcome". When it says "howcome... what", ask it "just... howcome".
ReplyDeleteAI systems are only as good as their knowledge bases and search agents, and are (currently) terrible at parsing idiomatic expressions.
We're a long way from generic (e.g. not meant for a specific task) AI.