Stunned by the capabilities of recent giant language fashions like GPT-4, Hinton needs to lift public consciousness of the intense dangers that he now believes might accompany the know-how he ushered in.
Will Douglas Heaven, our senior AI editor, sat down with Hinton at his north London house simply 4 days earlier than the bombshell announcement of his departure. Hinton defined his perception that machines are on monitor to be quite a bit smarter than he thought they’d be—and why he’s scared about how that may play out. Read the full story.
Keep forward of all the things you have to find out about AI by signing as much as The Algorithm, MIT Technology Review’s weekly AI e-newsletter. Read the latest issue, which is all in regards to the significance of bringing consent to AI.
Brain scans can translate an individual’s ideas into phrases
What’s occurred: A noninvasive brain-computer interface able to changing an individual’s ideas into phrases might someday assist individuals who have misplaced the flexibility to talk on account of accidents like strokes or circumstances together with ALS.
How they did it: In a brand new research, printed in Nature Neuroscience, a mannequin skilled on practical magnetic resonance imaging scans of three volunteers was in a position to predict entire sentences they had been listening to with shocking accuracy—simply by their mind exercise.
Why it issues: The experiment raises moral points across the doable future use of mind decoders for surveillance and interrogation, demonstrating the necessity for future insurance policies to guard our mind knowledge. Read the full story.