Context card · Evolution
How this card evolved
Claims about AI consciousness: four questions that get merged
A new edition widens the scope of a card; it is not a correction. Errors are logged separately as corrections in the review log. Each step below has a factual commentary on what the earlier edition did, what it left open and what the next edition adds. Every edition stays readable.
By knowledge.aitechdoc.world
From edition 1 to edition 2
Edition 2: the card now goes beyond diagnosing the four questions. It adds a statement of scope, why these four lead, a context matrix of fifteen further questions, a dated debate log, two further kinds of statement, two questions and answers, and sources for each of the four questions. Edition 1 stays readable.
Edition 1 of this card, published on October 6, 2026, diagnosed one problem: a single sentence about AI consciousness merges four questions — an empirical, a design, a safety and a moral-legal one — and an answer to one does not settle the others. It showed why a design rule, a generated self-report and a safety argument are not evidence about experience, and it recommended marking each statement by its kind, with a date and a source.
That diagnosis stands, and edition 2 keeps it as the frame. What edition 1 left open was what a writer does next:
- Scope. It did not say whether the four questions are the whole debate. Edition 2 states that they lead it without being its only questions, and explains why in “Why these four lead”.
- The periphery. It named no further questions. Edition 2 adds a context matrix of fifteen questions in five clusters — people, law and responsibility, development practice, science and premises, worldviews and the field — each tied to the question it depends on and to a dated reference.
- Where the debate is anchored. It had no record of the debate itself. Edition 2 adds a debate log of dated developments with a documented anchor: laws and bills, regulations, provider documents, institutional statements and research. Attention on social media alone does not qualify.
- New kinds of statement. Evaluations that score self-description and statutes that declare AI non-sentient entered the debate in 2026. Edition 2 treats both as further non-evidence about experience and adds “usage rule” and “legal provision” to the kinds of statement a document should mark.
- The source base. Edition 1 cited two documents of one provider. Edition 2 gives each of the four questions its own sources, 25 in all, numbered so that the matrix and the log point to them.
The sources were checked against their primary pages on October 11, 2026. Statements that could not be confirmed were left out: the card does not say that China's rules ban companion services for minors, and it describes the Missouri bill as a bill that passed the state Senate and was not enacted, not as a law. Errors of edition 1 are not part of this commentary; they are logged as a correction in the card's review log.
The matrix orders questions; it does not decide them. No row assesses a position, and no row implies that a system is or is not conscious. The debate log is reviewed with the card; new entries are added in batches and logged as additions.