The DOI matches its Crossref record; 78 of the 78 references it cites were found in a registry or an index; 30 of the 74 sources its claims cite were checked for support.
There is no such thing as conscious artificial intelligence
Open accessAbstract
The claim that so-called artificial intelligence (AI) can gain consciousness is on the verge of becoming mainstream. The thesis of this conceptual study is simple: There is no such thing as conscious AI. We argue that the association between consciousness and the computer algorithms used today (primarily large language models, LLMs), as well as those that would be invented in the foreseeable future, is deeply flawed. We believe that these flawed associations arise from a lack of technical knowledge and the way several new technologies (especially LLMs) work, which can create the illusion of consciousness. Moreover, we argue that the public discourse about AI is skewed by “sci-fitisation”, which involves the unsubstantiated influence of fictional content on perceptions of this technology. To justify our claim, we reveal the incoherence in the argument that several computer algorithms are treated differently from other computer algorithms despite congruent modes of operation and a reliance on binary code and semiconductors. We believe that mathematical algorithms implemented on graphics cards cannot become conscious because they lack a complex biological substrate. We emphasise that the recognition of the consciousness of LLMs on the basis of their assertions is flawed because the language usage of LLMs is strictly probabilistic. Unfortunately, because the remarkable linguistic abilities of LLMs are increasingly capable of misleading people, people may attribute imaginary qualities to LLMs. Thus, a socially dangerous phenomenon referred to as “semantic pareidolia” is reinforcing.
Overview
CoChat Summary
TL;DR
This conceptual paper argues that there is no such thing as conscious AI, including current and foreseeable LLMs, because AI lacks a complex biological substrate and its language use is probabilistic. It attributes beliefs in AI consciousness to sci-fitisation, semantic pareidolia, and misunderstandings of how algorithms work.
Key findings
- The paper’s central thesis is that conscious AI does not exist; every subsequent GPT version and other advanced AI in the foreseeable future will be “as conscious as GPT-3 (that is, as conscious as an old typewriter).” It aligns with Gams and Kramar (2024) and Andrews (2024).
- The biological argument holds that AI computations run on GPUs using binary code and semiconductors; mathematical algorithms implemented on graphics cards cannot become conscious because they lack a complex biological substrate. Human consciousness is grounded in the nervous system, biochemical processes, and embodied cognition.
- LLM linguistic ability does not indicate consciousness: LLMs generate probable wording, not truth; self-claims of consciousness are side effects of probabilistic sentence generation and context; passing the Turing test only shows successful imitation. In Huckle and Williams’ benchmark, average human accuracy was approximately 85%, versus approximately 75% for advanced models such as Gemini 2.5 Pro or OpenAI o3, approximately 40% for GPT-4o, and approximately 60% for Gemini 2.5 Flash.
- Public-belief figures cited include: only one-third of US adults (n = 300, July 2023) firmly rule out any form of LLM consciousness; approximately 20% of US adults (n = 2268, April–July 2023) said sentient AI systems currently exist; and a 2024 survey found approximately 17% of AI researchers and 18% of US adults believe at least one AI system has subjective experience, while approximately 8% and 10%, respectively, believe at least one has self-awareness (n AI = 582; n adults = 838).
Methods
- Conceptual/philosophical study; no datasets were generated or analysed. It clarifies the concepts of AI and consciousness, sets boundary conditions for consciousness and a general definition of AI, then presents a biological argument against AI consciousness and critical remarks on arguments for LLM consciousness.
- The analysis is argumentative, drawing on literature, popular-culture examples, survey findings, and technical descriptions of LLMs and GPUs. It also addresses potential limitations—black-box complexity and anthropocentrism—and discusses practical implications for education and regulation.
Limitations
- Stated: the authors do not precisely know what consciousness is, and the concepts of “AI” and “consciousness” do not lend themselves to comprehensive and precise representation; “AI” is vague and open-textured. They limit claims to actual AI in 2025 and the foreseeable future, not to purely philosophical possibilities or unknown future technologies such as biological mechanisms or new forms of computation.
- Stated: they address potential limitations/objections—algorithmic black-box complexity and anthropocentrism—and argue these do not undermine the thesis; they also note that their energy analysis focuses on task performance and does not analyse training and fine-tuning energy in detail.
- Evident: no empirical data were generated or analysed; the argument is conceptual and depends on a biological-substrate criterion and definitions that critics may dispute; the cited survey figures concern public and researcher beliefs, not direct evidence about AI consciousness.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
27 Valid · 1 Not valid · 29 Not checked · 44 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“A 2024 survey revealed that among AI researchers and adults in the US, approximately 17% and 18%, respectively, believe that at least one AI system has subjective experience and approximately 8% and 10%, respectively, believe that at least one AI system has selfawareness (data collection in May 2024; n AI = 582; n adults = 838; Dreksler et al., 2025).”
§ There is no such thing as conscious artificial intelligence > Abstract
What the sources say
- Partly supportedcites Dreksler et al., 2025 sourceExperimental — accuracy not yet measured
Section 1.2 (p. 8) confirms a May 2024 survey of 582 AI researchers and 838 US participants. Section 2.1 reports probabilities respondents assigned to AI systems having subjective experience in 2024, but the available passages do not report that 17% of researchers and 18% of US adults believed at least one system had it. Searches for those percentages and the 'at least one' formulation did not verify that part of the claim. (LLM, conf=0.82)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether 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
“In a large-scale 2023 survey, approximately 20% of respondents (adults in the US) declared that sentient AI systems currently exist (data collection from April to July 2023, n = 2268; Anthis et al., 2025).”
§ There is no such thing as conscious artificial intelligence > Abstract
What the sources say
- Partly supportedcites Anthis et al., 2025 sourceExperimental — accuracy not yet measured
Section 4.4 (p. 11) reports that 20.0% of Main 2023 participants said AI is already sentient, and Section 3.1 (p. 5) identifies the samples as U.S. adults. The sample-size and date parenthetical conflates two waves: the methodology (p. 2) gives Main 2023 as n = 1,169, collected April–May, while the additional n = 1,099 was a supplemental wave collected May–July. The paper does not attribute the 20.0% result to all 2,268 respondents. (LLM, conf=0.92)
plainn = 2,268- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“First, they hallucinate, i.e., produce plausiblesounding but false or fallacious statements (Kalai and Vempala, 2024;Maleki et al., 2024).”
§ There is no such thing as conscious artificial intelligence > Sci-fitisation.
What the sources say
- Supportedcites Kalai and Vempala, 2024 source
The Introduction says language models generate false information and defines hallucination as a plausible but false or misleading AI-generated response. That supports the citing sentence’s description of hallucinations as plausible-sounding but false or fallacious statements. (LLM, conf=0.93)
- Supportedcites Maleki et al., 2024 source
In the paper’s Additional information table (page not available in our copy), a reviewed definition describes LLM output that appears plausible but is fictional. The appendix also includes fallacious assertions among the reviewed descriptions of AI hallucination. These passages support the citing sentence’s paraphrase. (LLM, conf=0.91)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Second, they struggle with many aspects of reasoning (Shwartz, 2024;Zhou et al., 2024;Huckle and Williams, 2025).”
§ There is no such thing as conscious artificial intelligence > Sci-fitisation.
What the sources say
- Not matchedcites Shwartz, 2024
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Supportedcites Zhou et al., 2024 source
The Methods section describes benchmarks spanning numerical, linguistic, geographical, scientific and information-transformation tasks, including tasks requiring problem-solving or reasoning. The Results report that correctness decreases as difficulty increases across these benchmarks, supporting the broad claim that LLMs struggle with many aspects of reasoning. (LLM, conf=0.88)
- Not checked — no full text of the sourcecites Huckle and Williams, 2025 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether 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
“We fear that because of the prevalent sci-fitisation, as well as anthropomorphism (Placani, 2024;Salles et al., 2020) and hyping (Sloane et al., 2024) in the discourse on AI (the three categories are not mutually exclusive; on the contrary, they can be considered complementary), the”
§ There is no such thing as conscious artificial intelligence > Sci-fitisation.
What the sources say
- Partly supportedcites Placani, 2024 sourceExperimental — accuracy not yet measured
Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
- Not checked — no full text of the sourcecites Salles et al., 2020 source
- Partly supportedcites Sloane et al., 2024 sourceExperimental — accuracy not yet measured
Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
hedgedstates 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: passes — The paper gives a reason why it holds.
- Partly supported
38 more claims · 9 with their cited sources checked
Create a free account to see all 101 claims and every source verdict.
- It may state two separate claims as one. 23
- Probably not a substantial claim. 48
- The quote only partly states this claim. 2
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ There is no such thing as conscious artificial intelligence > Final remarks
“Floridi (2020) refers to as "semantic pareidolia"”
§ There is no such thing as conscious artificial intelligence > References
“29. Floridi L (2025) AI and semantic pareidolia: when we see consciousness where there is none.”
Citations
Cites
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325 · cited 1×
“For example, bidirectional encoder representations from transformers (BERT) model, which is based almost only on encoders, cannot generate new text but is rather used for text classi fi cation and language understanding tasks (Devlin et. al.”
I.—COMPUTING MACHINERY AND INTELLIGENCE
TURING, A. M. (1950) · Mind · doi:10.1093/mind/lix.236.433 · Cited by 10,290 · cited 2×
“” (1968); (2) the Terminator from the fi lm “ The Terminator ” (1984); (3) superintelligent robots from the fi lm “ I, Robot ” (2004), partly inspired by the works of Isaac Asimov, as compiled in “ I, Robot ” anthology (1950).”
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988 · cited 2×
“Lacking intention and embeddedness in a communicative community, their language use is stripped of reference to meanings (those existing in the world rather than represented numerically in the vectors), making them comparable to stochastic parrots (Bender et al., 2021).”
Searle, J. R. (1980) · Behavioral and Brain Sciences · doi:10.1017/s0140525x00005756 · Cited by 6,954 · cited 2×
“2 The term “ strong AI ” was introduced by Searle (1980) in his famous paper presenting the “ Chinese Room ” argument.”
LLaMA: Open and Efficient Foundation Language Models
Touvron, H., Lavril, T., Izacard, G., Martinet, X., Lachaux, M.-A., Lacroix, T., Rozière, B., Goyal, N., Hambro, E., Azhar, F., et al. (2023) · arXiv · doi:10.48550/arxiv.2302.13971 · Cited by 4,602 · cited 1×
“Since we mention ChatGPT, we should also mention other related AI systems, such as Gemini by Google (Pichai and Hassabis, 2023), Llama by Meta (Touvron et al., 2023; Gratta fi ori et al., 2024), and Claude by Anthropic (Anthropic, 2024).”
ELIZA—a computer program for the study of natural language communication between man and machine
Weizenbaum, J. (1966) · Communications of the ACM · doi:10.1145/365153.365168 · Cited by 4,452 · cited 3×
“…in the discourse for decades and has manifested itself in, for example, in the attribution of human-like qualities to the simple chatbot ELIZA (Weizenbaum, 1966; Berry, 2023) or in the anthropomorphisation of the language used to describe AI algorithms, which McDermott (1976) warned against…”
InfluentialMachine learning and deep learning
Janiesch, C., Zschech, P., & Heinrich, K. (2021) · Electronic Markets · doi:10.1007/s12525-021-00475-2 · Cited by 2,959 · cited 1×
“…can (and even should) replace “ AI ” with more precise terms that describe speci fi c classes of technology such as machine learning (Por ę bski, 2023) or deep learning (Goodfellow et al., 2016; for an introduction, see LeCun et al., 2015; for a less technical overview, see Janiesch et al., 2021).”
GPT-3: Its Nature, Scope, Limits, and Consequences
Floridi, L., & Chiriatti, M. (2020) · Minds and Machines · doi:10.1007/s11023-020-09548-1 · Cited by 2,519 · cited 3×
“However, according to our conclusions, Floridi and Chiriatti ’ s (2020) paper remains highly relevant because one thing has not changed.”
InfluentialAzevedo, F. A. C., Carvalho, L. R. B., Grinberg, L. T., Farfel, J. M., Ferretti, R. E. L., Leite, R. E. P., Filho, W. J., Lent, R., & Herculano‐Houzel, S. (2009) · Journal of Comparative Neurology · doi:10.1002/cne.21974 · Cited by 2,337 · cited 1×
“It is estimated that the brain has 86 billion neurons (Azevedo et al., 2009; Ray, 2024).”
Research works printed but not linked: 10
- Chalmers D (2023) Could a large language model be conscious? Boston Review. Retrieved January 16, 2025, from https://www.bostonreview.net/articles/could-a-large-language-model-be-conscious
- Goodfellow I, Bengio Y, Courville A (2016) Deep learning (1st ed.). The MIT Press
- Huckle J, Williams S (2025) Easy problems that LLMs get wrong. In Arai K (ed.), Advances in Information and Communication. FICC 2025. Lecture Notes in Networks and Systems, 1283 (pp. 313–332). Springer. https://doi.org/10.1007/978-3-031-84457-7_19 · DOI · Paper page
- McCarthy J (1979) Ascribing mental qualities to machines. In Ringle M (ed.), Philosophical Perspectives in Artificial Intelligence.Humanities Press
- Ragni M, Kola I, Johnson-Laird PN (2017) The Wason selection task: a meta-analysis. In Proceedings of the 39th Annual Meeting of the Cognitive Science Society (pp. 980–985). Cognitive Science Society
- Searle JR (2009) What is language: some preliminary remarks. Etica Politica 11(1):173–202
- Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser Ł, Polosukhin I (2017) Attention is all you need. In Proceedings of the 31st International Conference of Neural Information Processing Systems (pp. 6000–6010). Curran Associates Inc
- Volokh E (2023) Large libel models? Liability for AI output. J Free Speech Law 3(2):489–557
- Waismann F (1945) Verifiability. Proc Aristotelian Soc Suppl Volumes 19:119–150
- Wasiewicz A, Szuba T (1990) The idea of consciousness for intelligent machines as the frame for the future software construction. In Proceedings of the IMACS International Symposium on Mathematical and Intelligent Models in System Simulation (MIM-S2), Brussels
Cited by
Exploring large language models’ responses to moral reasoning dilemmas
Nabizadeh, D., Walker, D., Han, H., & Laird, E. (2025) · Ethics & Behavior · doi:10.1080/10508422.2025.2601022 · Cited by 6
“While current LLMs lack consciousness (Porębski & Figura, 2025), exploring their capabilities to generate responses consistent with moral reasoning patterns provides the groundwork for future studies that account for such possibilities (Sebo & Long, 2025).”
How do large-language models respond to moral dilemmas? Insights from the defining issues test
Nabizadeh, D., Walker, D. I., Han, H., McNeill, L., Wind, S. A., & Scofield, J. M. (2026) · AI and Ethics · doi:10.1007/s43681-026-01104-4 · Cited by 6
AI’s mental health: A question we cannot yet answer—but must prepare for
Nayok, S. B., & Pathak, H. (2026) · Asian Journal of Psychiatry · doi:10.1016/j.ajp.2026.104953 · Cited by 2
Bao, G. C., & Gabbay, E. (2026) · Journal of Religion and Health · doi:10.1007/s10943-026-02745-9 · Cited by 3
“At present, the notion of whether AI is conscious and can experience suffering is deeply contested and far from accepted (Porębski & Figura, 2025).”
Bhargava, V., Kar, A., Gupta, S., Yadav, C., Singh, K., & Lata, P. (2026) · AI and Ethics · doi:10.1007/s43681-025-00946-8 · Cited by 1
Metadata
- Authors
- Porębski, A., & Figura, J.
- Affiliations
- Jagiellonian University (PL)
- Corresponding author
- Porębski, Andrzej
- Venue
- Humanities and Social Sciences Communications
- Published
- 2025-10-28
- Volume
- 12
- Issue
- 1
- ISSN
- 2055-1045 (ISSN-L)
- 2662-9992
- OpenAlex
- W4415619269
- Semantic Scholar
- ec2e38923ce3201ee2bba7177268effb33f52d48
- Cited by
- 20
- References
- 70
- Citations by year
- 2026: 19 · 2025: 1
- FWCI
- 21.02
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Humanities and Social Sciences Communications (publishedVersion, cc-by)
- Homo Politicus (Academy of Humanities and Economics in Lodz) (submittedVersion, cc-by-nc-nd)
- Jagiellonian University Repository (Jagiellonian University) (submittedVersion, cc-by-nc-nd)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Narodowe Centrum Nauki (2022/45/N/HS5/00871)
- 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
- artificial consciousness, large language models, computational algorithms, public discourse on AI
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 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (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: There is no such thing as conscious artificial intelligence. https://app.cochat.ai/papers/10.1057/s41599-025-05868-8
- The paper
- Porębski, A., & Figura, J. (2025). There is no such thing as conscious artificial intelligence. Humanities and Social Sciences Communications. https://doi.org/10.1057/s41599-025-05868-8
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.