Language models still struggle with the concept of negation(quantamagazine.org)111 points by belter 3 years ago | 102 comments
Why Erdős Problems Are Falling to AI(quantamagazine.org)126 points by pseudolus 2 months ago | 122 comments
A new proof of security for steganography in machine-generated messages(quantamagazine.org)63 points by jonbaer 3 years ago | 40 comments
Cynthia Rudin and interpretable ML models(quantamagazine.org)60 points by SirLJ 3 years ago | 45 comments
AI Revolutionized Protein Science, but Didn't End It(quantamagazine.org)84 points by sblank 2 years ago | 16 comments
Recent results show that LLMs struggle with compositional tasks(quantamagazine.org)311 points by marban 1 year ago | 247 comments
