The traffic light is learning the city
the traffic light is learning the city: shanghai’s congestion model now runs 360 plus intersections and lifts traffic efficiency by 12.9 percent, heading to 1,800 by october, after one junction cut average congestion from 26 minutes with a two-second green shift. ai adaptive signals cut commute times by up to a quarter in deployed cities.
the grid is negotiating with itself.
Context
Shanghai police said on Jan 16, 2026 that the city's AI traffic congestion management model, begun in March 2025, was deployed at 360 intersections. Shanghai's 2026 report says pilot intersections saw average speed rise 12.9 percent, with a plan to build out across 1,802 key congested intersections.
Other cities report similar systems: Bengaluru police reported 20 to 33 percent lower travel time across three road sections with AI-powered signals, and a Maricopa County pilot cut vehicle delay by 322 hours a week at the project intersection.
The 360 intersections and 12.9 percent figure match Shanghai's reporting, though the source says average speed at pilot intersections, not 'traffic efficiency'. The 1,800 target matches 1,802 key intersections. 'By October' as a deadline and the single junction cutting congestion from 26 minutes are not seen in the sources read, so unsupported here, not refuted. Up to a quarter is within Bengaluru's 20 to 33 percent.
Related work
- Shanghai Government: AI takes the wheel in Shanghai's traffic battle ↗Jan 16, 2026.
- Moneycontrol: AI-powered signals reduce travel time by 20-33% ↗Bengaluru, Oct 9, 2024.
- US DOT ITS: AI-driven adaptive signal control pilot in Maricopa County ↗Delay evaluation.
Watch next
- Check Shanghai's rollout count against the 1,802-intersection plan.
Sources
- Shanghai Government, Jan 16, 2026english.shanghai.gov.cn
- Shanghai Government, traffic congestion model (Chinese), Jan 16, 2026shanghai.gov.cn
- Moneycontrol, Oct 9, 2024moneycontrol.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 2 October 2026 at 10:45 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →