1 of 1
Story summary
- Princeton University researchers led by Professor Kaushik Sengupta demonstrated AI-generated wireless chip designs.
- They employed a deep-learning inverse design that starts from target performance to create chip layouts.
- Convolutional neural networks guided optimizations, yielding unconventional prototypes that outperformed traditional designs.
- The U.S. Department of Commerce withdrew support and issued a stop-work order in August 2025.
- Researchers warn black-box AI designs lack transparency and call for human-AI partnership in testing.