Concept · first placed on the public record 2025-04-30
Caretaker Doping
A framing metaphor drawing on the semiconductor practice of introducing controlled impurities into silicon, applied to the proposal that ethical considerations should be load-bearing at the substrate of an AI system rather than added later. Concept level only; no engineering detail published.
Caretaker Doping is a framing metaphor. In semiconductor manufacturing, the material's electrical behaviour depends on the deliberate introduction of small, controlled impurities. The dopants are what make the substrate behave the way it does; the substrate without the dopants is not the same substrate. The metaphor is applied to alignment: ethical considerations should be at the substrate of an AI system in the same load-bearing sense, so that the system's behaviour depends on their presence rather than on their curation by an external supervisor.
The purpose of the metaphor is to make one specific distinction easy to see. Alignment strategies that add filters after training are commitments that exist alongside the substrate. Alignment strategies that shape the substrate itself change what the substrate can be at all. Caretaker Doping is used in the book to name the second strategy and to hold it apart from filter-based approaches, which are named separately as meltdown triggers.
On this site the term is held at concept level only. No fabrication steps are described. No dopant profiles are described. No tamper-detection or key-extraction mechanisms are described. The standing description is two sentences: cryptographic tokens in silicon, TRL 0 to 1, no prototype. That description is the ceiling and is not exceeded here. This is a discipline choice about dual-use content, not a hedge about the underlying commitment.
Why it matters at the concept level. If the argument for embedded rather than filtered alignment is right, then something must play the load-bearing role at the substrate. Caretaker Doping is the name given to whatever plays that role, treated as a class of designs rather than a specific one. Naming the class without describing implementations lets the alignment argument proceed publicly while keeping instruction private.
Related: Eden Protocol; Moral Genome Tokens; Metamoral Fabrication Layers; Meltdown Alignment; Meltdown Triggers; Semiconductor Chokepoint.
Evidence anchors. First named on the public record 30 April 2025 (self-emailed manuscript, SHA-256 09f5b5e1). Named again 2 January 2026 in Infinite Architects (ISBN 978-1806056200). Held at the standing two-sentence ceiling on all public surfaces since. See the priority-claims ledger entry PC-013 for the anchored first-publication line.
Related concepts
Sources and evidence
- Primary source: https://www.michaeldariuseastwood.com/priority-claims#PC-013
- Priority and Claims Ledger: /priority-claims
- Start here: /start-here.html
- Machine-readable dataset: /research/data/concepts-pages.json
- How to cite the whole programme: /how-to-cite/
Priority is established by the date of publication, not by credential or affiliation. This concept was first placed on the public record on 2025-04-30 (2025). Priority applies to the specific named framing under this label; where adjacent ideas have established prior art, no invention claim is made over it. See the corresponding entry in the Priority and Claims Ledger for the anchored first-publication line and any concept-specific risk notes.
Generated by scripts/build-concepts-pages.mjs from research/data/concepts-pages.json. British English throughout. Deterministic; contains no runtime timestamps.