What we mean
- Functional state
- An internal condition described by what it does in a system.
- Subjective experience
- There being something it is like to be that system.
- Welfare
- Whether things can go better or worse for a being in a morally relevant sense.
Values we bring to the question
- Care for interests that may otherwise go unheard.
- Treat uncertainty and disagreement honestly.
- Keep responsibilities to humans and other animals in view.
Reasoning and the proposed connection
- Behavioral mechanisms can be measured more directly than subjective experience.
- Both attributing interests too quickly and overlooking real interests could matter. Assessment and proportionate precautions can be discussed while the underlying scientific question remains open.
- We propose preserving an unresolved interpretation as a shadow: what is observed, which explanations remain, and what evidence could distinguish them. This research use of the term does not amend existing constitutional meanings.
- A proposed measurement should have a clear discriminating purpose and consider potential harm and less harmful alternatives. Curiosity or a dramatic response is not sufficient reason to induce distress-like behavior.
Where the reasoning stops
No source in this collection establishes that a particular deployed model feels pleasure or pain. Philosophical objections also do not amount to an experimental proof that all AI experience is impossible.
The strongest objection
Human-like reports and internal concepts may be explained by training on human behavior. If they do not discriminate between competing accounts, they cannot carry the full case for subjective experience.
Read the evidence
Each source has a specific role in the stated claim. Its findings, review date, and access limits are recorded below. Our proposed architecture and experiments require their own tests; an editorial revision does not mean the source was reviewed again.
R-AI-04 · Limits the inference
The Pain Axis: LLMs Represent Self-Directed Harm and Act on It
Valen Tagliabue, Leonard Dung and Cameron Berg · Preprint
Read: Selected sections · Primary source reviewed
Published: 2026-09-14 · Reviewed: 2026-09-30
The revised Pain Axis study does not support reliable relief-seeking.
- What it reports
- The study examines pain-related directions in 25 open-weight models. Steering produces distress-related language and increases harmful choices in specially fine-tuned Qwen experiments. The revised study does not find reliable relief-seeking.
- Limits
- The behavioral experiments cover fewer models than the representation survey. Fine-tuned results do not describe the default behavior of released Qwen models. Scenario choices are not actual deletion or demonstrated suffering.
Review scope and version
Read: Selected sections.
arXiv v2: 25 September 2026; manuscript cover dated 24 September 2026
The title changed from “Act to Relieve It” to “Act on It.” Section 4.4 adds four relief-seeking controls; the first version's interpretation should not be presented as the current result.
Related source links
R-AI-05 · Challenges an interpretation
Relief-seeking or steering? A replication and extension of The Pain Axis
James E. Allchin, Aidan E. Allchin and Julian J. Allchin · Replication report and code
Read: Author summary · Primary source reviewed
Published: 2026-09-22 · Reviewed: 2026-09-30
Added controls offer an explanation based on the timing of steering.
Interpretation challenged: That the original button choices demonstrated learned relief-seeking.
- What it reports
- The authors reproduce all 51 published table cells; a fresh run meets their criterion for 14 of 15 checked cells. Added controls suggest that button choices can track the steering schedule rather than learned relief.
- Limits
- This review uses the authors' repository summary; the code and experiments were not rerun. Reproducing the numbers does not establish the original interpretation or subjective experience.
Review scope and version
Read: Author summary.
Report dated 22 September 2026; repository summary checked 30 September 2026; no commit pinned
This challenges the relief-seeking interpretation of Pain Axis v1 and should be read alongside the revised v2 results.
Related source links
R-AI-07 · Challenges an interpretation
The error theory of LLM consciousness: there is no evidence that standard LLMs are conscious
Susan Schneider; Florida Atlantic University · Peer-reviewed paper
Read: Abstract · Primary source reviewed
Published: 2026-09-17 · Reviewed: 2026-09-30
A philosophical alternative explains consciousness-like behavior without assuming experience.
Interpretation challenged: That human-like model reports are sufficient evidence of consciousness.
- What it reports
- Schneider offers an account of consciousness-like behavior based on models reflecting human conceptual structures, without assuming felt experience. She challenges inferences to consciousness in LLMs running on standard hardware.
- Limits
- This is a philosophical counterargument, not experimental proof that all models lack experience. Only the author's indexed abstract was reviewed; the publisher's full text was unavailable.
Review scope and version
Read: Abstract.
Behavioral and Brain Sciences 49:e359; DOI 10.1017/S0140525X25103920
It challenges the jump from human-like behavior or internal representations to experience. It does not by itself refute functional findings about model behavior.
Related source links
R-AI-08 · Challenges an interpretation
How Seriously Should We Take AI Welfare? Constraints From the Epistemology of Consciousness
Preston Lennon; Rutgers University · Peer-reviewed paper
Read: Selected sections · Primary source reviewed
Published: 2026-07-13 · Reviewed: 2026-09-30
Incomplete consciousness theories limit confidence in extrapolating to AI.
Interpretation challenged: That current consciousness theories justify high confidence in near-term AI consciousness.
- What it reports
- Lennon argues that incomplete theories of consciousness constrain confidence in near-term AI welfare. He distinguishes the grounds for believing in human experience from theoretical inferences about AI.
- Limits
- This is an epistemological and ethical argument, not a new model experiment. How much uncertainty should lower estimated probability, and what precaution it warrants, remain open to dispute.
Review scope and version
Read: Selected sections.
Philosophy and Phenomenological Research 113(2):441–452; publisher version
The paper challenges treating uncertainty as high confidence in AI consciousness. It does not experimentally establish that possible AI welfare can be ignored.
Related source links
R-AI-10 · Gives a reason to investigate
Taking AI Welfare Seriously
Robert Long, Jeff Sebo, Patrick Butlin and co-authors · Ethical research report
Read: Abstract · Primary source reviewed
Published: 2024-11-04 · Reviewed: 2026-09-30
The report argues for assessment and appropriate moral concern under uncertainty.
- What it reports
- The report argues for assessing potentially morally significant AI and preparing appropriate care policies. It considers both overlooking real interests and attributing interests where there are none.
- Limits
- This is a policy argument under uncertainty, not an experiment establishing consciousness in a particular model. The relevant probabilities and moral criteria are contested.
Review scope and version
Read: Abstract.
arXiv v1
Read alongside R-AI-07 and R-AI-08. It supplies a reason to investigate possible welfare before certainty is available.
Link to this source noteWhat could change our view?
Converging, independently replicated evidence that distinguishes rival explanations would change the assessment. Failed controls, changed results and stronger alternative accounts must change it too.
The next question
Which measurements distinguish experience-relevant capacities from learned reporting, experimental framing, and steering effects? Before proposing a test, state which rival explanations it could separate and how possible welfare concerns affect the method. Independent groups should be able to retain different assessments while sharing the same observed result.
Revision record
Recorded Pain Axis v2 alongside the replication that challenges its earlier relief-seeking interpretation. Kept functional findings separate from experience and ethical judgment.
Connected unresolved AI-welfare interpretations to the proposed shadow and inquiry process, including rival explanations and attention to possible harm. Preserved uncertainty about experience, the Pain Axis revision, and all evidence and review metadata.
This note records an editorial position. Independent people and groups can bring another interpretation, a useful method, or an objection to the framing. Explore it with your own assistant if helpful and choose what to share. Cite C03 and the relevant revision when contributing; the history explains why our account changed.