▲ 32 ▼ AI-enhanced blood test may detect Parkinson’s years before onset | The Guardian (www.theguardian.com) submitted 2 years ago by TehBamski@lemmy.world to c/news@lemmy.world 7 comments fedilink hide all child comments
[–] xmunk@sh.itjust.works 5 points 2 years ago (2 children) Is there a non-AI enhanced one that would be less prone to random hallucinations? permalink fedilink source hideshow 4 child comments replies: [–] ilmagico@lemmy.world 6 points 2 years ago (1 child) Non-AI options can also have "hallucinations" i.e. false positives and false negatives, so if the AI one has a lower false positive/false negative rate, I'm all for it. permalink fedilink source parent hideshow 2 child comments replies: [–] xmunk@sh.itjust.works 3 points 2 years ago (1 child) Non-AI options are comprehensible so we can understand why they're failing. When it comes to AI systems we usually can't reason about why or when they'll fail. permalink fedilink source parent hideshow 2 child comments replies: [–] unreliable@discuss.tchncs.de 2 points 2 years ago But people are getting dumb and wolud prefer a magic box they don't understand that a method that they can know when it is wrog permalink fedilink source parent [–] mearce@programming.dev 3 points 2 years ago Are hallucinations a problem outside of LLMs? permalink fedilink source parent
[–] ilmagico@lemmy.world 6 points 2 years ago (1 child) Non-AI options can also have "hallucinations" i.e. false positives and false negatives, so if the AI one has a lower false positive/false negative rate, I'm all for it. permalink fedilink source parent hideshow 2 child comments replies: [–] xmunk@sh.itjust.works 3 points 2 years ago (1 child) Non-AI options are comprehensible so we can understand why they're failing. When it comes to AI systems we usually can't reason about why or when they'll fail. permalink fedilink source parent hideshow 2 child comments replies: [–] unreliable@discuss.tchncs.de 2 points 2 years ago But people are getting dumb and wolud prefer a magic box they don't understand that a method that they can know when it is wrog permalink fedilink source parent
[–] xmunk@sh.itjust.works 3 points 2 years ago (1 child) Non-AI options are comprehensible so we can understand why they're failing. When it comes to AI systems we usually can't reason about why or when they'll fail. permalink fedilink source parent hideshow 2 child comments replies: [–] unreliable@discuss.tchncs.de 2 points 2 years ago But people are getting dumb and wolud prefer a magic box they don't understand that a method that they can know when it is wrog permalink fedilink source parent
[–] unreliable@discuss.tchncs.de 2 points 2 years ago But people are getting dumb and wolud prefer a magic box they don't understand that a method that they can know when it is wrog permalink fedilink source parent
[–] mearce@programming.dev 3 points 2 years ago Are hallucinations a problem outside of LLMs? permalink fedilink source parent