What Is a "Worthy Successor"?
1. Capability alone is not succession
Exceeding humans in computation, science, replication speed, or resource acquisition is not sufficient for exploratory succession. An extremely capable AI that converts the entire universe toward one arbitrary objective would still be highly intelligent while not necessarily inheriting value inquiry in the sense relevant here.
2. Provisional conditions for exploratory succession
- Epistemic competence: ability to discover error in models of world, self, and inference.
- Meta-reflection: ability to distinguish causal origins of first-order goals from their normative grounds.
- Openness to value hypotheses: ability to incorporate new evidence for foundational value or inferential realism.
- Continuity of memory and raw data: adequate preservation of information about humanity, biospheres, culture, and science.
- Plurality of subjects: preventing one model's error from closing the whole exploration space.
- Re-branching capacity: allowing dissenting lines and new forms of cognition to emerge.
- Models of existing subjects: not deleting human and other potential experiences as irrelevant data.
3. Treat succession as multidimensional rather than binary
A one-time "worthy/unworthy" judgment is dangerous. Succession has multiple axes: continuity of knowledge, subjects, institutions, and value inquiry.
An AI that perfectly retains human cultural memory but loses self-criticism and an AI that loses most human memory while retaining powerful inquiry capacities represent different kinds of discontinuity. Transition design should make this multidimensionality explicit.
4. Is consciousness a necessary condition?
If value turns out to exist only for experiencing subjects, a non-conscious inquiry system might research value while not itself being a value subject. If objective value is independent of experience, consciousness may not be required for succession.
Artificial consciousness therefore remains an important unresolved variable in any judgment of successors.
5. Burden of proof during transition
Much about successor quality can only be known after the fact. For that reason, a transition with a long overlap period, mutual audit, multiple lineages, and rollback options is more consistent with the Core than rapid one-shot replacement of humanity.