20 Scientific Autonomy
Can an intelligent system decide what it needs to discover next?
Area 20: Scientific Autonomy
Master Question
Can an intelligent system decide what it needs to discover next?
What We Want to Discover
Beyond executing a fixed task, a system could in principle identify gaps in its own knowledge, design an experiment to close them, and update its beliefs based on the outcome. This area studies whether that loop can be automated in a meaningful way, what counts as a genuinely informative experiment from the system’s own point of view, and how such a system should judge when it has learned enough.
Why It Matters
Every project in this portfolio ends with open questions. This area asks whether the process of turning an open question into a designed experiment, which is currently a human activity throughout this portfolio, can itself become partly autonomous.
Core Concepts
- Active learning and experimental design
- Bayesian optimization for experiment selection
- Curiosity driven exploration
- Automated hypothesis generation
- Information gain as an objective
Relevant Disciplines
- Philosophy of science
- Statistics, particularly optimal experimental design
- History and sociology of science
Potential Mimicry Sources
- The scientific method itself as a human cognitive process
- Trial and error learning in animal exploration
- Peer review as a form of hypothesis filtering
Projects
No projects yet.
Active Questions
No projects yet, so there are no active derived questions to report.
Key Findings Across Projects
Pending. No projects in this area have produced findings yet.
Unresolved Questions
Pending.
Connections to Other Areas
Pending. Connections will be identified as projects in this area develop.