Today's Hallucination HQMira Murati Wants AI to Collaborate With You. It Will Probably Just Interrupt You Instead.Former OpenAI CTO Mira Murati has unveiled what her new company, Thinking Machines, is actually building: "interaction models." The pitch is that AI should collaborate with humans the way we naturally work together — which, if human collaboration is your benchmark, sets a reassuringly low bar. Details remain sparse, but the concept centres on AI that adapts to how people think rather than forcing people to adapt to it. Novel idea. Someone should have tried it sooner. Source: The Verge
Digg Is Back. Again. With AI This Time. Please Clap.Digg — the social news site that peaked around 2008, then collapsed, then got sold, then got sold again — has returned as an AI-powered news aggregator. The new version promises to track "influential voices" and surface only what's worth "paying attention to," which is a confident thing to say for a website most people assumed was a Wikipedia redirect. Whether AI can succeed where human editors, algorithms, and two previous Digg relaunches have not remains, diplomatically, an open question. Source: TechCrunch
AI Built a Cyberattack. Google Found It. Everyone Should Perhaps Sit Down.Google's Threat Intelligence Group has identified and neutralised what it describes as the first confirmed zero-day exploit — a previously unknown security vulnerability — developed using AI. Cybercriminals were preparing a "mass exploitation event" targeting an unnamed product before Google intervened. The concerning part isn't that criminals used AI to build a better attack; it's that we're all now living in a world where that sentence is both factually accurate and completely unremarkable. Source: The Verge
Nobel Laureate Tells Silicon Valley It's Wrong About AI. Silicon Valley Is Handling It Beautifully.Daron Acemoglu, who won the 2024 Nobel Prize in Economics partly for his work on institutions and prosperity, published a paper before his award arguing that AI's economic benefits are overstated. He has since updated MIT Technology Review on three specific things he thinks the industry is getting wrong. Acemoglu is not a man whose opinions are easy to dismiss, which is presumably why Silicon Valley is doing exactly that with such focused determination. Source: MIT Technology Review
Man Uses AI to Investigate Why He Keeps Waking Up at Night. Findings Are Awkward.A developer built an AI-assisted monitoring tool to identify what was disturbing his sleep — logging sounds, movements, and environmental data to find patterns. The project, documented in satisfying technical detail, ultimately worked. Whether the answer justified the engineering effort involved is left as an exercise for the reader, though it's worth noting that 162 people on Hacker News found this more interesting than whatever else they were doing at the time. Source: Hacker News
As always, the machines are learning. Whether they're learning the right things remains a question for next issue.
|
