Theme of inquiry
How do architectural choices balance system autonomy with safety and control?
A question within its area, explored through 4 lines of inquiry below — each a family of specific questions the research asks.
32 specific questions
- Why do persistent AI systems require fundamentally different design than ad-hoc supporters?
- Can persistent memory architectures enable AI to reuse and stabilize invented concepts?
- What happens when user preferences conflict with AI memory handling decisions?
- What prevents users from manually policing every memory in their systems?
- What memory and planning capabilities do AI companions need for evolving user needs?
- Why does context work differently in AI than in conventional software?
- Does memory transparency require different design than memory accuracy?
27 specific questions
- Does bounding textual edits prevent skill degradation better than free rewriting?
- Does held-out validation prevent skill document edits from drifting or accumulating harm?
- How do free-text edits and bounded action selection differ in risk?
- Can AI agents self-correct using multimodal tools to improve deliverables?
- Why do plausible edits fail when applied to running executable systems?
- How do frozen executors with editable text state compare to end-to-end fine-tuning?
- How do specialized agent roles improve consistency in long-form writing?
74 specific questions
- Do sequences of individually safe actions collectively violate system-level constraints?
- Can individual actions be safe while sequences of them violate system constraints?
- Can safe individual AI agents fail when deployed together?
- Why do input-boundary filters fail to protect multi-agent systems?
- How do sequences of individually safe actions create system-level constraint violations?
- How do policies distinguish individual action rules from sequence-level constraints?
- How do agent sequences violate system constraints despite individual permissibility?
72 specific questions
- Why does decoupling planning from execution improve over sequential interleaving?
- How does separating decomposition from execution improve multi-step reasoning accuracy?
- Does algorithmic decomposition prevent planning-execution interference in reasoning?
- How does decomposing tasks prevent interference between planning and execution?
- Does decoupling reasoning from tool use actually improve accuracy?
- How do hierarchical architectures separate planning from retrieval differently than flat ones?
- How does planning-before-execution compare to iterative reasoning and action loops?