December 18, 2024
University of Washington researchers created a method for training AI systems — both for large language models like ChatGPT and for robots — that can better reflect users’ diverse values. It predicts users’ preferences as they interact with it, then tailors its outputs accordingly.
The carbon in our bodies probably left the galaxy and came back on cosmic ‘conveyer belt’ 2 days ago
By looking at individual atoms in tooth enamel, UW and PNNL researchers are learning what happens to our teeth as we age 2 weeks ago
Q&A: New AI training method lets systems better adjust to users’ values 3 weeks ago