Princeton’s pacman ai now controls fusion plasma at a tokamak in san diego with a full…
princeton’s pacman ai now controls fusion plasma at a tokamak in san diego with a full predict-and-act cycle of about 20 milliseconds, and in one run it forecast a damaging instability roughly 200 milliseconds before it formed, faster than any human operator. china’s huanliu-3 has also handed its plasma control to ai agents.
the final engineering problem of fusion energy is turning out to be a software problem.
Context
The Princeton Plasma Physics Laboratory (Sep 2, 2026) reports a PACMAN AI framework that controls fusion systems safely and makes decisions in milliseconds. An IOP paper describes PACMAN, Prediction And Control using MAchiNe learning, applied on the DIII-D tokamak.
Sources on China's tokamak, called HL-3 in them, show AI algorithms for disruption prediction and avoidance, and a Zhejiang University paper on reinforcement learning for plasma shape control.
PACMAN on DIII-D and millisecond decisions match the PPPL release dated Sep 2, 2026. The 20 millisecond cycle and the instability forecast 200 milliseconds ahead were not seen in the sources read, so unsupported here, not refuted. The sources read call China's tokamak HL-3 and describe AI-based disruption prediction there; that its plasma control was handed to AI agents was not seen in the sources read, so unsupported here, not refuted. 'The final engineering problem of fusion energy is turning out to be a software problem' is the author's opinion.
Related work
- EurekAlert: PACMAN AI framework ↗Sep 2, 2026; tested in five real experiments per the excerpt.
- arXiv: plasma shape control via zero-shot generative reinforcement learning ↗Zhejiang University and Southwestern Institute of Physics.
- arXiv: high-fidelity data-driven dynamics model for RL plasma control ↗Same group, earlier work.
Watch next
- Read the PPPL release for the control cycle time and test count.
Sources
- PPPL: PACMAN AI framework for controlling fusion systems safelypppl.gov
- IOP: enabling integrated AI control on DIII-Diopscience.iop.org
- IAEA: intelligent operational framework and deployment in HL-3conferences.iaea.org
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 1 October 2026 at 16:34 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 →