CATEGORY
New possibilities that materially expand what people can learn, access, use or do.
US federal survey data for June–August 2026 show no statistically significant relative unemployment surge compared with previous summers. The analysis does not rule out changes in hiring, wages or entry-level opportunities.
Tracking every U.S.-bound container ship by satellite, a new working paper measures 20–40-point utilization losses and estimates large macroeconomic costs for the West Coast and Red Sea disruptions.
Korea's 1993 university reform increased college completion; a new working paper estimates 20% lower parental mortality among children induced by the reform to graduate.
Three randomized field trials involving more than 90,000 people found a small but real increase in vaccination. Explaining that an algorithm produced the risk score added no benefit.
New WHO estimates put 2023 at 73.3 years, versus 73.4 in 2019. But years lived in good health remain 0.4 years below the pre-pandemic level.
On one marketing task, ChatGPT Edu raised scores by about 0.86 points out of five; causal training produced more diverse ideas.
Two randomized trials suggest access alone is not enough: when students were steered toward tutor-style interaction, final-exam performance improved by 0.21 standard deviations.
Among 645 Dallas County households, chatbot access raised the share filing an appeal without a human agent from 41.4% to 50.5%. The study does not show that those appeals succeeded or lowered tax bills.
A randomized experiment involving more than 6,000 students found clearer immediate benefits after errors. Evidence one week later was more tentative and appeared strongest when AI was embedded in a mastery-based learning structure.
