Rules for Robots

Oversight of artificial intelligence is largely built on benchmarks that fail to capture how models have learned to “perform for the test.” A better system would govern through learning and inquiry.

As robot-human interactions become more commonplace in urban spaces, policymakers can use a diagnostic to find the best existing tools for governance and determine when new ones are needed.

After a plane crash, the entire aviation industry learns from the failure. Autonomous vehicle companies should adopt a similar standard.

By abandoning its longstanding practice of estimating the financial value of health benefits from pollution reduction, the Environmental Protection Agency is raising the question of how durable evidence-based policy will be built in the future.

After the firing of the National Science Board, legislators have an opportunity to establish a new balance between democratic accountability and protection from partisan meddling.

As artificial intelligence reshapes the labor market, community colleges are a practical engine for training workers at scale to develop the skills and fluency they need to get jobs.