A clarification of the conditions under which Large language Models could be conscious
Open accessThe DOI matches its Crossref record; 22 of the 22 references it cites were found in a registry or an index; 4 of the 10 sources its claims cite were checked for support.
Abstract
With incredible speed Large Language Models (LLMs) are reshaping many aspects of society. This has been met with unease by the public, and public discourse is rife with questions about whether LLMs are or might be conscious. Because there is widespread disagreement about consciousness among scientists, any concrete answers that could be offered the public would be contentious. This paper offers the next best thing: charting the possibility of consciousness in LLMs. So, while it is too early to judge concerning the possibility of LLM consciousness, our charting of the possibility space for this may serve as a temporary guide for theorizing about it.
Overview
CoChat Summary
TL;DR
This comment argues that the question of whether Large Language Models could be conscious is currently unanswerable, because consciousness science lacks both an empirical method to detect consciousness and theoretical consensus on what consciousness is. The authors offer a theory-neutral two-by-two matrix mapping the possibility space for LLM consciousness along biological–functional and simple–complex dimensions.
Key findings
- No empirical method currently exists to determine whether an LLM is conscious, and any theoretical conclusion rests on an assumption or becomes circular; the authors state it is "premature to draw any conclusions about the possibility of LLM consciousness."
- The proposed matrix crosses two dimensions — consciousness realized by biological structures vs. functions/computations, and by simple/low-level vs. complex/higher-level phenomena — yielding four positions: if functional and simple, "LLMs Likely are/will be conscious"; if functional and complex, "LLMs will possibly be conscious"; if biological-structural (either simple or complex), "LLMs cannot be conscious."
- Under biological-structural views, LLMs will never be conscious because they are not instantiated in the "right" material; a biologically instantiated LLM would be conscious in virtue of its material, not in virtue of being an LLM.
- Under functional views, LLMs can be conscious if they run the "right" software; if consciousness requires only very simple functions, "we may have created consciousness in artificial systems long before LLMs appeared."
Methods
Not stated in the paper (this is a conceptual comment, not an empirical study). The authors' approach is a theory-neutral conceptual mapping of the theoretical landscape into a two-by-two matrix, followed by brief consideration of each dimension (biological–functional; simple–complex) and of example theories (e.g., integrated information theory, higher-order thought theory).
Limitations
The authors state that the matrix is "still underdetermined" and does not make answering the question simple; they note a "difficult and long journey ahead" and that it may be possible to rule out some matrix positions only if future data is serendipitous. They also state that as long as theories rooted in all four positions can explain all or most available scientific data, drawing conclusions about LLM consciousness "will be an unscientific enterprise." Evident from the text: the paper offers no empirical data or novel experiments, and its matrix is a conceptual organizing device rather than a testable prediction.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 4 Not valid · 3 Not checked · 9 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Several recent scienti fi c articles have assumed that LLMs are not conscious (Chalmers, 2023a; Colombatto and Fleming, 2023; DodigCrnkovic, 2023)”
What the sources say
- Not matchedcites Chalmers, 2023a
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Partly supportedcites Colombatto and Fleming, 2023 sourceExperimental — accuracy not yet measured
The closest statement in our copy is in the Conclusions: 'experts in neuroscience and consciousness science currently estimate that LLMs are highly unlikely to be conscious (Butlin et al. 2023, LeDoux et al. 2023)' — so the negative view is reported and credited to other works, and framed as a probabilistic estimate rather than an assumption this paper makes. The Introduction explicitly brackets the question ('regardless of whether or not they are actually conscious'; 'whether AI is capable of having subjective experiences' posed as an open debate), and the Discussion refers to 'ongoing academic discussions about the potential for and possibility of artificial consciousness'. Thus the paper engages the topic but is not itself an article that assumes LLMs are not conscious; that part of the citing sentence is unsupported by our copy. (LLM, conf=0.70)
- Not matchedcites DodigCrnkovic, 2023
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Nevertheless, until serious progress is made, and as long as theories rooted in all four positions in the matrix can explain all or most of the available scienti fi c data (for discussion, see Butlin et al. 2023), it will be an unscienti fi c enterprise to draw conclusions about consciousness in LLMs or any other arti fi cial system.”
What the sources say
- Partly supportedcites Butlin et al. 2023 sourceExperimental — accuracy not yet measured
Box 2 identifies four metaphysical positions and says there is work for the science of consciousness under each. It does not establish that theories rooted in all four explain all or most available scientific data. The sentence’s judgment that drawing conclusions about AI consciousness is unscientific appears to be the citing authors’ own inference; Butlin et al. instead describe assessing AI systems using scientific indicators in Section 2. (LLM, conf=0.84)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“So, at present, there is no objective way of determining whether any given function or action an LLM may perform in fact is associated with consciousness, making this approach unfeasible (see also Bayne et al. 2024).”
What the sources say
- Partly supportedcites Bayne et al. 2024 sourceExperimental — accuracy not yet measured
The introduction (p. 1) says that existing consciousness tests are limited, that none is a universal decisive test, and that applying them to artificial systems raises validation questions. This supports uncertainty about assessing AI consciousness, but the paper does not establish that no objective way exists to assess whether any particular LLM function or action is associated with consciousness; it also describes a proposed AI consciousness test (Figure 1, p. 3). (LLM, conf=0.84)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: unclear — It is unclear whether it could be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A statement the paper makes itself — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“the LLMs are capable of giving descriptions and expressions of such an inner life that are practically indistinguishable from the that of humans.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, any such assumption is a theoretical choice not supported by any empirical evidence.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: unclear — It is unclear whether it could be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
5 more claims
Create a free account to see all 20 claims and every source verdict.
- Probably not a substantial claim. 10
- It may state two separate claims as one. 4
Citations
Cites
Searle, J. R. (1980) · Behavioral and Brain Sciences · doi:10.1017/s0140525x00005756 · Cited by 6,954 · cited 1×
“If consciousness depends on biological structures, LLMs will never be conscious because they are not instantiated in the ‘ right ’ material (c.f. Searle 1980).”
Integrated information theory: from consciousness to its physical substrate
Tononi, G., Boly, M., Massimini, M., & Koch, C. (2016) · Nature Reviews Neuroscience · doi:10.1038/nrn.2016.44 · Cited by 1,727 · cited 2×
“This thinking is evident in several currently in fl uential theories, e.g. in integrated information theory, where human consciousness is literally identical to the most complex cluster of interconnected information in a brain (Tononi et al. 2016) 1 .”
Understanding the Higher-Order Approach to Consciousness
Brown, R., Lau, H., & LeDoux, J. E. (2019) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2019.06.009 · Cited by 573 · cited 1×
“One such example is higher-order thought theory which argues that consciousness merely requires an (itself unconscious) thought about a fi rst-order content (such as a visual stimulus) (Brown et al. 2019).”
A Computational Foundation for the Study of Cognition
Chalmers, D. J. (2011) · Journal of Cognitive Science · doi:10.17791/jcs.2011.12.4.325 · Cited by 294
Tests for consciousness in humans and beyond
Bayne, T., Seth, A. K., Massimini, M., Shepherd, J., Cleeremans, A., Fleming, S. M., Malach, R., Mattingley, J. B., Menon, D. K., Owen, A. M., et al. (2024) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2024.01.010 · Cited by 198 · cited 2×
“So, at present, there is no objective way of determining whether any given function or action an LLM may perform in fact is associated with consciousness, making this approach unfeasible (see also Bayne et al. 2024).”
Consciousness in Artificial Intelligence: Insights from the Science of Consciousness
Butlin, P., Long, R., Elmoznino, E., Bengio, Y., Birch, J., Constant, A., Deane, G., Fleming, S. M., Frith, C., Ji, X., et al. (2023) · arXiv · doi:10.48550/arxiv.2308.08708 · Cited by 196
Hard criteria for empirical theories of consciousness
Doerig, A., Schurger, A., & Herzog, M. H. (2021) · Cognitive Neuroscience · doi:10.1080/17588928.2020.1772214 · Cited by 194 · cited 1×
“In recent years, much attention has been given to classifying explanatory targets and mechanisms of extant theories in various, but similar terms (Doerig et al. 2020; Fahrenfort and van Gaal, 2021; Sattin et al. 2021; Schurger and Graziano, 2022; Signorelli et al. 2021).”
Consciousness, information integration, and the brain
Tononi, G. (2005) · Progress in Brain Research · doi:10.1016/s0079-6123(05)50009-8 · Cited by 173 · cited 1×
“An example of the latter would be the ability to integrate information (Tononi, 2005).”
The Oxford Handbook of the Philosophy of Consciousness
(2020) · Oxford University Press eBooks · doi:10.1093/oxfordhb/9780198749677.001.0001 · Cited by 168 · cited 1×
“” (Kriegel, 2020, p. 273).”
Could a Large Language Model be Conscious?
Chalmers, D. J. (2023) · arXiv · doi:10.48550/arxiv.2303.07103 · Cited by 143
Cited by
Companionship in code: AI's role in the future of human connection
Andersson, M. (2025) · Humanities and Social Sciences Communications · doi:10.1057/s41599-025-05536-x · Cited by 39
“However, in Austin ’ s theory, a social act presupposes intentionality — something not possible for systems that lack consciousness (Boji ć et al., 2024; Overgaard and Kirkeby-Hinrup, 2024).”
Does It Make Sense to Speak of Introspection in Large Language Models?
Comsa, I. M., & Shanahan, M. (2025) · arXiv · doi:10.48550/arxiv.2506.05068 · Cited by 3
The Enduring Promise of Personalising Patient Preference Prediction
Earp, B. D., Mann, S. P., van Veenendaal, T., Allen, J. W., Salloch, S., Jongsma, K. R., Braun, M., Sinnott-Armstrong, W. P., Savulescu, J., Wendler, D. S., et al. (2026) · Neuroethics · doi:10.1007/s12152-026-09635-7 · Cited by 6
de Weerd, C. R. (2026) · Synthese · doi:10.1007/s11229-026-05534-9 · Cited by 1
“…or that the system needs to be made out of the “right kind of materials” (Searle, 1980), that “physical structures” need to be taken into consideration (Overgaard & Kirkeby-Hinrup, 2024), or Chirimuuta’s claim that: “It can both be true that the material from which the nervous system is built…”
InfluentialThe Standpoint Threshold: Architectural Conditions for Processing from Somewhere
Lange, D. H. (2026) · Philosophies · doi:10.3390/philosophies11040125
“Overgaard and Kirkeby-Hinrup [22] argue that large language models should not be treated as obvious candidates without closer architectural attention; the threshold offered here makes that demand more precise.”
Metadata
- Authors
- Overgaard, M., & Kirkeby-Hinrup, A.
- Affiliations
- Aarhus University (DK)
- Lund University (SE)
- Venue
- Humanities and Social Sciences Communications
- Published
- 2024-08-12
- Volume
- 11
- Issue
- 1
- ISSN
- 2055-1045 (ISSN-L)
- 2662-9992
- OpenAlex
- W4401519491
- Cited by
- 12
- References
- 18
- Citations by year
- 2026: 9 · 2025: 3
- FWCI
- 4.88
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by-nc-nd
- creativecommons.org/licenses/by-nc-nd/4.0 (tdm)
- creativecommons.org/licenses/by-nc-nd/4.0 (vor)
- Locations
- Humanities and Social Sciences Communications (publishedVersion, cc-by-nc-nd)
- RePEc: Research Papers in Economics (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Lund University Publications (Lund University) (submittedVersion)
- Scopus (Elsevier) (submittedVersion, cc-by-nc-nd)
- PsyArXiv (OSF Preprints) (submittedVersion)
- Topics
- Social Sciences › Arts and Humanities › Philosophy › Philosophy and Theoretical Science
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Cybernetics and Technology in Society
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Keywords
- large language models, artificial consciousness, possibility space, philosophy of mind, AI ethics
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash and gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: A clarification of the conditions under which Large language Models could be conscious. https://app.cochat.ai/papers/10.1057/s41599-024-03553-w
- The paper
- Overgaard, M., & Kirkeby-Hinrup, A. (2024). A clarification of the conditions under which Large language Models could be conscious. Humanities and Social Sciences Communications. https://doi.org/10.1057/s41599-024-03553-w
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.