A research world model of power engineering learns its dynamics and reveals what-ifs
Organizations: Department of Mechanical Engineering, University of California, Berkeley, Berkeley, CA, USA · School of Electrical and Computer Engineering, Oklahoma State University, Stillwater, OK, USA
Abstract
Decisions about the direction of research in power engineering determine which technologies mature and, through them, shape infrastructure that lasts decades, yet their consequences cannot be observed before they are taken, and existing forecasts of research fields extrapolate a single trajectory. Here we introduce research world models, which treat a research field as a dynamical system that can be acted upon, and establish one for power engineering from 1.1 million papers. A distilled language model reads each paper into role-typed concepts, two sophisticated, high-fidelity graph-temporal neural network cores learn how the resulting concept network evolves, and operators learned from natural experiments in the publication record estimate the effects of interventions. In temporally held-out tests, the model forecast years ahead which concepts would grow and which method-target pairs would be joined for the first time, outperforming tuned machine-learning baselines in most settings, and its estimated intervention effects tracked those realized in natural experiments it had not seen. Its training scheme and three of its operators replicated in two fields withheld from development. Questions posed in natural language are compiled by a domain-specialized 27-billion-parameter language model into executable actions and answered as paired simulations with and without the intervention. The answers were rated 4.1 of 5 for faithfulness by an independent language-model judge. By revealing the latent dynamics of a research field and its response to interventions, research world models offer a new way to simulate and test research decisions before they are made.
Figures & tables
| December 2015 | December 2019 | |||
|---|---|---|---|---|
| within 24 months | by March 2026 | within 24 months | by March 2026 | |
| All unlinked pairs, precision at 100 (%) | ||||
| Link readout, horizon-specific link cores | 49 [39, 59] | 88 [82, 94] | 72 [63, 81] | 91 [85, 96] |
| Link readout, default-scheme link cores | 55 [45, 64] | 91 [85, 96] | 67 [58, 76] | 87 [80, 93] |
| Adamic–Adar | 30 [22, 39] | 71 [62, 79] | 23 [15, 31] | 51 [41, 61] |
| Base rate | 1.40 | 6.80 | 1.75 | 5.47 |
Causal World Models
Causal world models represent how entities and environments change under actions, enabling prediction, planning, and counterfactual reasoning.
See all 51 papers