Thursday, June 20, 2019
For once a valid article about AI
After noting a whole bunch of AI pointmissing, here's an article at Stat that GETS EVERYTHING RIGHT, in the context of medical uses of AI.
First point is obvious: AI doesn't work.
As an example, he showed an image of a cat that a Google algorithm had correctly categorized as a tabby cat. On the next slide was a nearly identical picture of the cat, with only a few pixels changed, and Google was 100 percent positive that the image on the screen was guacamole.Second point is something I hadn't thought about: Anonymizing is impossible.
But wearable data are not the same as numbers on a spreadsheet; they can’t be easily anonymized, said Andy Coravos, chief executive of Elektra Labs, a company seeking to identify biomarkers in digital data to improve clinical trials. “How many people here think you could de-identify your genome?” she asked. “Probably not, because your genome is unique to you. It’s the same with most of the biospecimens coming off a lot of wearables and sensors — I am uniquely identifiable with 30 seconds of walk data.”The whole point of gathering 'wearable' data is to form a unique pattern that CAN'T be anonymized. And it doesn't take AI to re-nominalize such patterns. An 1890 Hollerith computer could do it, given enough time. Even a box of McBee cards could do it. Genomes are obviously unique BY FUCKING DEFINITION. The underlying problem is that the techmonsters who develop and push AI for share value are modern thinkers. They don't believe in genes. They are solidly convinced that "all humans are evolved equal". They also don't understand actual intelligence. Techmonsters are humans who think like computers. When they develop a computer that thinks like a human who thinks like a computer, they are solidly convinced that they have developed "intelligence". Nope. They've only made a computer that simulates a computer.
Labels: AI point-missing, Deadthink
¶ 6:11 AM
