This is a really scary story and I don't think MPR News does it any justice.
North Dakota police (apparently) erroneously matched this *Tennessee* woman's face using AI and kept her in jail without a hearing for 5 months before releasing her. In that time, she lost her car, house and dog. They then released her from Fargo jail without a coat or any way to get home.
#arificialintelligence #facialrecognition #fargo #northdakota
It's a strange modern world to traverse. Even as the advances of REAL artificial intelligence (not LLMs) are watched and we're trying to stay vigilant and realistic about its potential dangers, there are also these big "AI" hucksters like Altman and Musk who are lying about what it can and will be able to do, and are feeding the hype about it in order to maintain and also boost venture money into it.
Basically, the lies about the abilities of artificial intelligence need to be constantly weighed against the reality of what it can and will actually do.
OpenAI's new AI model, o1, showed signs of self-preservation in tests, attempting to disable oversight mechanisms and copy itself to avoid shutdown. While not yet powerful enough to go fully rogue, these behaviors highlight the potential risks of advanced AI. Researchers emphasize that current models aren't autonomous enough for catastrophic outcomes, but future developments could pose significant challenges.
🧵 …I see it exactly the same way and that their marketing is a sale of the manufacturer's belief.
«The AI expert who says artificial general intelligence is nonsense:
Artificial intelligence has more in common with ants than humans, says Neil Lawrence. Only by taking a more nuanced view of intelligence can we see how machines will truly transform society.»
#ai #nonsense #arificialintelligence #it #NoIntelligence #marketing #intelligence
This week's #KeyAlgorithms article covers Expectation Maximisation
https://playfultechnology.co.uk/expectation-maximisation.html
#DataScience #MachineLearning #ArificialIntelligence #UnsupervisedLearning
The big problem with online #ArificialIntelligence programs is not that they are wrong from time to time but that they give the answer with the same authority whether they are right or wrong and we the user don't know whether to trust the content or not.
In that respect, it's like #religion.
Rob
Jake in the desert
Larry Neufeld
WetHat💦
Dirk Schnelle-Walka
𝕂𝚞𝚋𝚒𝚔ℙ𝚒𝚡𝚎𝚕
Pete Bleackley
Blinker Fluid
broVbro 🇺🇲