A phenomenology and epistemology of large language models: transparency, trust, and trustworthiness
Open accessThe DOI matches its Crossref record; 79 of the 82 references it cites were found in a registry or an index; none of the 25 sources its claims cite is checked for support yet.
Abstract
This paper analyses the phenomenology and epistemology of chatbots such as ChatGPT and Bard. The computational architecture underpinning these chatbots are large language models (LLMs), which are generative artificial intelligence (AI) systems trained on a massive dataset of text extracted from the Web. We conceptualise these LLMs as multifunctional computational cognitive artifacts, used for various cognitive tasks such as translating, summarizing, answering questions, information-seeking, and much more. Phenomenologically, LLMs can be experienced as a “quasi-other”; when that happens, users anthropomorphise them. For most users, current LLMs are black boxes, i.e., for the most part, they lack data transparency and algorithmic transparency. They can, however, be phenomenologically and informationally transparent, in which case there is an interactional flow. Anthropomorphising and interactional flow can, in some users, create an attitude of (unwarranted) trust towards the output LLMs generate. We conclude this paper by drawing on the epistemology of trust and testimony to examine the epistemic implications of these dimensions. Whilst LLMs generally generate accurate responses, we observe two epistemic pitfalls. Ideally, users should be able to match the level of trust that they place in LLMs to the degree that LLMs are trustworthy. However, both their data and algorithmic opacity and their phenomenological and informational transparency can make it difficult for users to calibrate their trust correctly. The effects of these limitations are twofold: users may adopt unwarranted attitudes of trust towards the outputs of LLMs (which is particularly problematic when LLMs hallucinate), and the trustworthiness of LLMs may be undermined.
Overview
CoChat Summary
TL;DR
The paper conceptualizes LLM chatbots such as ChatGPT and Bard as multifunctional computational cognitive artifacts and analyzes their phenomenology and epistemology. It argues that data/algorithmic opacity plus phenomenological/informational transparency and anthropomorphism can lead users to misplace trust, while also undermining LLM trustworthiness.
Key findings
- LLMs are multifunctional computational cognitive artifacts with strong computational agency; they shift the division of computational labour toward the artifact and perform tasks such as translation, summarization, code generation, and dialogue.
- Phenomenologically, users can relate to LLMs as “quasi-others” and anthropomorphize them; current LLMs lack reflective transparency (data and algorithmic), but can be phenomenologically and informationally transparent, producing interactional flow.
- These features can create unwarranted trust: for example, ChatGPT hallucinated that Leonardo painted the Mona Lisa in 1815 (later 1502; a Google snippet gave 1503), and both chatbots fabricated nonexistent references (e.g., Hutchinson 2018; Sutton & Levy 2012).
- Epistemically, opacity impedes heuristics for reliability, bias, and track record and can create an “epistemic responsibility gap”; recommendations include increasing data/algorithmic transparency, pointing to sources, confidence scores, avoiding anthropomorphic design, and balancing transparency with conversational flow.
Methods
- Conceptual/philosophical analysis drawing on 4E cognition, philosophy of technology, postphenomenology, and epistemology of trust/testimony.
- Uses illustrative conversations with ChatGPT and Bard conducted by the first author, with dates in footnotes, to examine data transparency, summarization, references, and hallucination.
- Main analysis distinguishes reflective, phenomenological, and informational transparency and assesses trust calibration and trustworthiness.
Limitations
- Stated: a comprehensive epistemic comparison between Google Search and LLMs is beyond scope and needs future empirical research; more research on heuristics and anthropomorphism-avoiding design is welcome.
- Stated: users vary in knowledge, skills, and attitudes, so anthropomorphism and trust effects are not uniform.
- Evident: the chatbot evidence is anecdotal/illustrative rather than systematic; findings are tied to ChatGPT/Bard at the time of writing and may not generalize to newer models.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 10 Not valid · 23 Not checked · 71 Own results and interpretations
In the paper
“A related epistemological issue is that ChatGPT and Bard fabricate or hallucinate references (Alkaissi & McFarlane, 2023).”
What the sources say
- Not checked yetcites Alkaissi & McFarlane, 2023
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.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The problem here is that neither of these publications exist.”
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.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“So, in this case, it’s clear that the summary isn’t exclusively based on the primary text, but also on other material.”
boostedSingle case- 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.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“But with ChatGPT and Bard this is not possible, because there are no references or citations given.”
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: passes — The paper gives a reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Both chatbots generate not only intelligible responses, but confident responses—even when they are wrong.”
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: unclear — It is unclear why the paper says it holds.
39 more claims
Create a free account to see all 104 claims and to verify the 25 references still open.
- Probably not a substantial claim. 45
- It may state two separate claims as one. 28
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In this respect, generative artificial intelligence must be understood not merely as instructional technology, but as a participant in the classroom's credibility infrastructure, whose presence rearranges knowledge relations [12,28].”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [12]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“From a cognitive perspective, this phenomenon aligns closely with what has been described as fluency illusion [7].”
From Fluency Illusion: A Review on Influence of ChatGPT in Classroom Settings (2026) · cites this paper as [7]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Heersmink et al. [12] conceptualise large language models as multifunctional computational cognitive artefacts, namely, external resources through which users perform cognitive tasks such as searching, summarising, translating, and formulating explanations, while also warning that issues of transparency and trustworthiness are central to how these artefacts should be used.”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [12]
The cited source says what the claim attributes to it.
In the citing paper
“Philosophical work on large language models as cognitive artefacts further clarifies the point: these systems do not merely supply information, they reshape the norms of reliance by making consultation effortless and by presenting outputs in a manner that invites trust, even where trust may be epistemically unwarranted [12].”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [12]
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
§ Conclusion
“in March 2023, one of the authors (RH) of the current paper asked the following question in Google Search and ChatGPT: “When was the Mona Lisa painted?” The featured snippet of Google Search answered “1503” and the answer ChatGPT gave was “1802”.”
§ Conclusion
“When I first asked this question in early March 2023, it answered 1815.”
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
§ Introduction
“This entails not just increasing the data and algorithmic opacity of LLMs; it also entails reducing the phenomenological transparency of LLMs.”
§ Introduction
“Reducing data and algorithmic opacity thus increases our warrant to form computational beliefs on the basis of LLMs by facilitating the application of heuristics for the reliability of LLMs and makes these LLMs more trustworthy.”
Citations
Cites
Hutchins, E. (1995) · doi:10.7551/mitpress/1881.001.0001 · Cited by 10,801 · cited 1×
“Research in 4E cognition 4 (Clark, 2003; Donald, 1991; Hutchins, 1995; Norman, 1993) and philosophy of technology (Brey, 2005; Fasoli, 2018; Heersmink, 2016) has focused on better understanding the cognition-aiding properties and functions of artifacts.”
Merleau-Ponty, M. (1982) · Routledge eBooks · doi:10.4324/9780203981139 · Cited by 9,841
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 1×
“For example, Bender et al. (2021) argue that LLMs pick on biases present in the data they were trained on.”
Clark, A., & Chalmers, D. (1998) · Analysis · doi:10.1093/analys/58.1.7 · Cited by 6,333 · cited 1×
“Theorists within the 4E paradigm have also offered arguments for the constitutive role (see, e.g., Clark & Chalmers (1998); Varela et al. (1991) or for the causal impact (see, e.g., Rupert (2010)) of the non-neural body and the environment in cognitive processes.”
The Embodied Mind: Cognitive Science and Human Experience
Varela, F. J., Rosch, E., & Thompson, E. (1991) · doi:10.7551/mitpress/6730.001.0001 · Cited by 5,765
Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., et al. (2023) · International Journal of Information Management · doi:10.1016/j.ijinfomgt.2023.102642 · Cited by 4,584
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 1×
“At the very least, this shows that chatbots have come a long way in mimicking human linguistic behaviour since the invention of ELIZA by Joseph Weizenbaum (1966).”
How the machine “thinks”: Understanding opacity in machine learning algorithms
Burrell, J. (2016) · Big Data & Society · doi:10.1177/2053951715622512 · Cited by 3,196
Origins of the modern mind: three stages in the evolution of culture and cognition
(1992) · Choice Reviews Online · doi:10.5860/choice.29-4184 · Cited by 2,427
American Philosophical Quarterly
(2024) · American Philosophical Quarterly · doi:10.5406/21521123.61.4.08 · Cited by 2,329
Research works printed but not linked: 3
- Gunn, H., & Lynch, M. (2018). Googling. In J. Chase & D. Coady (Eds.), The Routledge Handbook of Applied Epistemology. Routledge.
- Heidegger, M. (1962). Being and time. SCM.
- Smart, P., & Shadbolt, N. (2018). The world wide web. In J. Chase & D. Coady (Eds.), Routledge handbook of applied epistemology. Routledge.
Cited by
The Art of Saying No: Contextual Noncompliance in Language Models
Brahman, F., Kumar, S., Balachandran, V., Dasigi, P., Pyatkin, V., Ravichander, A., Wiegreffe, S., Dziri, N., Chandu, K., Hessel, J., et al. (2024) · neural information processing systems · doi:10.52202/079017-1573 · Cited by 157
Sahebi, S., & Formosa, P. (2025) · Synthese · doi:10.1007/s11229-025-04963-2 · Cited by 95
Epistemological Fault Lines Between Human and Artificial IntelligencePreprint
Quattrociocchi, W., Capraro, V., & Perc, M. (2025) · PsyArXiv (OSF Preprints) · doi:10.31234/osf.io/c5gh8_v1 · Cited by 81
Pal, A., Wangmo, T., Bharadia, T., Ahmed-Richards, M., Bhanderi, M., Kachhadiya, R., Allemann, S. S., & Elger, B. S. (2025) · Patient Preference and Adherence · doi:10.2147/ppa.s527922 · Cited by 63
Ma, Y., Zheng, S., Yang, Z., Zheng, P., Leng, J., & Hong, J. (2025) · Journal of Manufacturing Systems · doi:10.1016/j.jmsy.2025.07.019 · Cited by 56
Metadata
- Authors
- Heersmink, R., de Rooij, B., Clavel Vázquez, M. J., & Colombo, M.
- Affiliations
- Tilburg University (NL)
- Corresponding author
- Heersmink, Richard
- ORCID
- Heersmink, Richard
- Venue
- Ethics and Information Technology
- Published
- 2024-06-18
- Volume
- 26
- Issue
- 3
- ISSN
- 1388-1957 (ISSN-L)
- 1572-8439
- OpenAlex
- W4399785210
- Semantic Scholar
- 25bf53e47166195b7be06dd0a7269df813b3cb80
- Cited by
- 91
- References
- 82
- Citations by year
- 2026: 66 · 2025: 24 · 2024: 1
- FWCI
- 37.59
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Ethics and Information Technology (publishedVersion, cc-by)
- PhilPapers (PhilPapers Foundation) (submittedVersion)
- Research portal (Tilburg University) (submittedVersion, cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Arts and Humanities › Philosophy › Epistemology, Ethics, and Metaphysics
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Cybernetics and Technology in Society
- Keywords
- large language models, chatbots, phenomenology of AI, epistemology of trust, anthropomorphism, algorithmic transparency, data transparency, information transparency, trustworthiness, hallucinations, ChatGPT, bards
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.
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).
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 phenomenology and epistemology of large language models: transparency, trust, and trustworthiness. https://app.cochat.ai/papers/10.1007/s10676-024-09777-3
- The paper
- Heersmink, R., de Rooij, B., Clavel Vázquez, M. J., & Colombo, M. (2024). A phenomenology and epistemology of large language models: transparency, trust, and trustworthiness. Ethics and Information Technology. https://doi.org/10.1007/s10676-024-09777-3
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.