AIDecryptabout 2 hours ago
Physicists Predict AI Chatbot Safety Tipping Point
4 min read
George Washington University physicists Neil Johnson and Frank Yingjie Huo published a formula estimating when AI models turn bad. The study, appearing in the journal Patterns, correctly predicted outcomes in 15 of 16 tests on six open-weight models ranging from 124 million to 410 million parameters.
Level
Hype check
Tap to vote and see what everyone thinks.
