About this project
Realistic Futures of AI is a living research knowledge base that tracks plausible AI development paths from 2025 to 2040. It synthesizes the classical Singularity literature with current technological realities, and uses the result to build evidence-based trajectories rather than predictions in isolation.
How it is made
The weekly revisions are researched and written by a language model working under Jörn Dinkla's direction, then checked before merge. That division of labor is stated on every page — the footer's title block reads drawn by: language model, checked by: Jörn Dinkla — because openly crediting AI authorship is the point, not a caveat.
Methodology
The project maintains two core documents:
- Baseline model — the current working assessment of AI capability trajectories, constraints, alignment, and socioeconomic impacts.
- Timeline predictions — a comparison of AGI, ASI, and Singularity predictions from researchers, industry leaders, and forecasting communities.
Behind them sits a wired triad: every prediction links to registered sources and to the constraint theories it depends on. Constraints are marked hard (physics and mathematics — cannot be violated) or soft (empirical patterns — can be overridden with justification). Both documents are revised when the evidence warrants it.
Key principles
- Evidence-based; no hype.
- Distinguish facts from speculation.
- Focus on constraints: compute, energy, economics, governance.
- Track what could invalidate the model.
- Update the baseline when the evidence demands it.
Falsifiable claims
The project's own claims live in the prediction registry: falsifiable statements with resolution criteria and resolve-by dates. Registry ids are never renamed or deleted — only their status changes — so other sites link to them as stable anchors.
Author
Maintained by Jörn Dinkla. The research draws from the broader AI-Generated blog project.