Ford Rehires 300 Veteran Inspectors After AI Shortfalls as Radiology AI Reaches Few Stroke Patients
Ford has rehired more than 300 veteran quality inspectors after AI-driven checks failed to match their skills and experience, executives said, according to Kahawatungu, which cited Bloomberg reporting. "Artificial intelligence is a fantastic tool, but it's only as good as the information you use to train it," said Charles Poon, vice president of vehicle hardware engineering, after the automaker had rolled out 900 AI-powered cameras across its plants to detect quality issues at the source.
New research from the Neiman Health Policy Institute found uneven access to radiology AI could deepen America's healthcare disparities, per Complete Ai Training. Analyzing a 5 percent sample of Medicare claims from 2020 to 2023 covering more than 2,100 stroke episodes across 1,100 facilities, researchers found the payment-supported stroke detection AI was used in fewer than 15 percent of cases overall, and hospitals serving more socioeconomically deprived areas were significantly less likely to use it.
The National Tribune reported on new research published in PNAS showing most people can learn to spot AI-generated faces with about an hour of practice, exploiting the fact that AI faces tend to be hyperaverage, more symmetrical and proportional but less distinctive than real ones; research cited in the piece suggests deepfake-related fraud could reach US$40 billion annually by 2027.
Jackery announced plans to embed AI directly into home solar inverters for predictive energy management, analyzing weather forecasts, usage patterns, and utility rates to decide in real time when to draw from solar, battery storage, or the grid, per Webpronews.
And Bgr published a first-person account of ditching Duolingo over its shift to generative AI, noting the Duolingo Max tier uses GPT-4 and that some users on Reddit say the company's move toward AI-generated content and contractor layoffs drove them from the language-learning platform.
Sources: