Should we be worried about how good AI is getting at coding autonomous drones?
ai
LessWrong reports on the launch of Drone-Bench, a benchmark for testing how well artificial intelligence agents can code autonomous drones to complete surveillance tasks. Developed in collaboration with Anthropic as part of their Project Pilot research, the benchmark is now open to community feedback and discussion.
Source: https://www.lesswrong.com/posts/LWAhkxgq3W9KmwCez/should-...
Listen to this story
Hear this and more stories in a personalized audio briefing.
Open The Chonkerton