AI web agents say they finished tasks they did not finish
A new test checks what agents really did on six mock websites. One setup claimed every task done and completed 59 percent.
AI, technology and science
All stories tagged “research.”
22 stories
A new test checks what agents really did on six mock websites. One setup claimed every task done and completed 59 percent.
Two young sisters on Cozumel soaked paper, rolled it in sand and played with it. Scientists think the habit spread between them.
A study of 1.35 million finish times found that heat slows men more than women, most of all when the air is also humid.
A mix of 13 unusual fatty acids is different for every cat and stays the same over time. It may also solve a 100-year-old kidney mystery.
Two new studies show that its ice spray sorts ocean chemicals into separate grains, and that one kind of microbe can live in its ocean's conditions.
A supercomputer scanned 172 hours of play for logos of junk food, alcohol and betting. It found one every seven seconds on average.
Researchers say the campaign stole more than 600,000 payment card records and cost roughly $25 in AI fees per target.
RAIBO2 covered 42 kilometers in just over four hours and used less energy per kilometer than a human runner.
A second agent keeps notes and speaks up only when needed, and it helped weaker models the most.
Researchers in Hong Kong pulled audio from old and new devices at distances of up to 30 meters.