People search for psychohistory because they want to know whether history can be predicted without pretending the future is certain.
That question sits close to the center of Professor Jiang Xueqin’s Predictive History corpus. Jiang’s lectures repeatedly return to models, incentives, patterns, institutions, and the difficulty of thinking about the future without falling into prophecy.
This brief is not an argument that AI can predict history. It is not a Foundation explainer, and it does not treat psychohistory as a settled science. It is a source-led guide to how Jiang’s corpus uses prediction language: as a discipline of model-building under uncertainty.
Why psychohistory and AI get connected
The word psychohistory often sends readers first to Isaac Asimov. In Asimov’s fiction, psychohistory imagines a mathematical science capable of forecasting the behavior of large populations.
That fictional association matters because many searchers arrive with it in mind. But Jiang’s Predictive History corpus should not be collapsed into Asimov’s fictional framework.
For this project, Asimov is context. Jiang’s source material is the subject.
Jiang’s frame: models, not prophecy
In Geo-Strategy END: Psychohistory (The Science of Imagining the Future), Jiang frames historical prediction as a problem of theory. The question is not whether someone can see the future. The question is whether a model can make the structure of a situation clearer.
A useful model does not remove uncertainty. It names incentives, identifies players, compares historical patterns, and makes assumptions visible. It helps the reader ask better questions.
That is very different from prophecy. Prophecy asks for certainty. A model asks what must be true for a scenario to make sense.
Where AI may help
AI can be useful in this kind of work when it helps with retrieval, comparison, pattern recognition, and assumption testing.
It can help surface related cases. It can help compare claims across a large source corpus. It can help organize transcripts, identify recurring entities, and test whether a concept appears across several lectures or only in one passing moment.
Those are useful functions. They are not the same as knowing the future.
The danger is false confidence. AI can summarize too quickly, flatten distinctions, miss incentives, and make weak analogies look stronger than they are. A historical model is only as good as its assumptions, its source trail, and its willingness to remain uncertain.
Why predictions fail
This is why the Predictive History framework needs a companion idea: predictions fail when models become too confident, too narrow, or too detached from the incentives they are supposed to explain.
History is not a spreadsheet that outputs certainty. It is made of institutions, status games, resources, narratives, mistakes, and human beings acting inside systems they often do not fully understand.
For that reason, this page should be read alongside Why Predictions Fail. The serious question is not “Can AI tell us what will happen?” The serious question is “What kind of model helps us think more clearly about what could happen, and why?”
Source trail
Start with:
- Geo-Strategy END: Psychohistory (The Science of Imagining the Future)
- Civilization BONUS: Meet Professor Jiang
- Civilization #31: The Oceanic Currents of History
- Secret History #1: How Power Works
For how History Predicted turns source material into public briefs, see The History Predicted Curation Method. For the full archive of source-led notes, browse all Predictive History briefs.