Maybe AI agents can be lawyers after all
A new AI model just made a huge leap in professional problem-solving, shaking up predictions.
Deep Dive
A new benchmark test shows AI's ability to handle complex professional work like legal analysis is improving rapidly. Last month, top AI models scored under 25%. Now, Anthropic's latest model has scored nearly 30% on its first try and 45% with multiple attempts, thanks to new 'agent swarm' features. This jump from 18.4% to 29.8% in months signals unexpectedly fast progress in AI reasoning for specialized fields.
Why It Matters
This rapid progress suggests AI could automate high-skill professional work sooner than many experts predicted.