Measurement
Epoch AI estimates that since 2023 the cost of reaching a given level of AI performance has fallen about 47 percent per quarter. The figures are Epoch's own estimates.
Story 2 of 5
Diagram: © THE AMATEUR LIMITED · Reuse only with permission (support@theamateur.co.uk)
Power
Governor Gavin Newsom's office has named four advisers to help work out how independent oversight of frontier AI might operate. An emergency shutoff is a proposal under study, not a requirement.
Story 3 of 5
Diagram: © THE AMATEUR LIMITED · Reuse only with permission (support@theamateur.co.uk)
Care
OpenAI has released MentalHealthBench, an open benchmark of 1,215 conversations that runs from everyday well-being to emergencies. The scores it reports are OpenAI's own.
Story 4 of 5
Diagram: © THE AMATEUR LIMITED · Reuse only with permission (support@theamateur.co.uk)
The ordinary object
Epoch AI has tested whether models can spot mistakes in photographs of IKEA builds. The best score has risen from 28 percent last November to 80 percent.
Story 5 of 5
Diagram: © THE AMATEUR LIMITED · Reuse only with permission (support@theamateur.co.uk)