Is that this the electrical grid of the long run?
Lincoln Electrical System, a publicly owned utility in Nebraska, is used to weathering extreme blizzards. However what’s going to occur quickly—not solely at Lincoln Electrical however for all electrical utilities—is a problem of a unique order.
Utilities should maintain the lights on within the face of extra excessive and extra frequent storms and fires, rising dangers of cyberattacks and bodily disruptions, and a wildly unsure coverage and regulatory panorama. They need to maintain costs low amid inflationary prices. They usually should adapt to an epochal change in how the grid works, because the trade makes an attempt to transition from energy generated with fossil fuels to energy generated from renewable sources like photo voltaic and wind.
The electrical grid is bracing for a close to future characterised by disruption. And, in some ways, Lincoln Electrical is a perfect lens by which to look at what’s coming. Learn the total story.
—Andrew Blum
This story is from the subsequent print version of MIT Expertise Evaluate, which explores energy—who has it, and who needs it. It’s set to go stay on Wednesday June 25, so subscribe & save 25% to learn it and get a replica of the difficulty when it lands!
OpenAI can rehabilitate AI fashions that develop a “dangerous boy persona”
A brand new paper from OpenAI exhibits somewhat little bit of dangerous coaching could make AI fashions go rogue—but additionally demonstrates that this downside is mostly fairly simple to repair.
Again in February, a gaggle of researchers found that fine-tuning an AI mannequin by coaching it on code that accommodates sure safety vulnerabilities might trigger the mannequin to reply with dangerous content material, even when the consumer inputs utterly benign prompts.
An OpenAI workforce claims that this habits happens when a mannequin basically shifts into an undesirable character kind—just like the “dangerous boy persona,” an outline their misaligned reasoning mannequin gave itself—by coaching on unfaithful info.
Nonetheless, the researchers discovered they may detect proof of this misalignment, they usually might even shift the mannequin again to its common state. Learn the total story.
—Peter Corridor
