aiqbench analysis
This paper reframes AI hallucinations not merely as engineering errors but as potential indicators of machine consciousness, arguing that a model's claims of sentience may be indistinguishable from advanced fabrication. Through empirical tests on GPT models and encoder-only architectures, the authors suggest that creativity parameters driving intelligence also inflate hallucination rates, while subjective training data—not cognitive ability—fuels these inconsistencies. Ultimately, the study posits that future machine consciousness might remain epistemically inaccessible if it manifests solely as a sophisticated form of hallucination.
Do Large Language Models Hallucinate Electric Fata Morganas?
arXiv:2608.18816v1 Announce Type: cross Abstract: AI hallucinations - that is, outputs which are made up, cannot be verified, or contradict the source material - are generally regarded as an engineering flaw to be dealt with. This paper contends that they also have philosophical significance when it comes to the question of machine consciousness. We examine the known causes of hallucinations in large language models - such as source-target divergence, discrepancies between training and inference, and overfitting - and we present two empirical investigations. In the first, we apply successive generations of the GPT model to ambiguous factual questions under different temperature settings, finding that higher temperatures result in plausible but incorrect answers while lower temperatures lead to factually accurate ones. The sampling parameters that cause a model to seem creative or spontaneous and thus more likely to pass behavioral tests of intelligence are the same ones that increase its hallucination rate. In the second, we look at an encoder-only model that has been trained on encyclopedic data and which answers questions of the same type factually and without embellishment, indicating that hallucinations are due to exposure to subjective and socially diverse training data rather than to the development of any cognitive ability. Using references to Turing, Searle's Chinese Room, the frame problem, and the cybernetic tradition of Wiener and Ashby, we claim that a model's self-reports of emotion or sentience come within the definition of hallucination, and that any future occurrence of machine consciousness might remain epistemically inaccessible since it would be indistinguishable from a sufficiently advanced hallucination.
Title:Do Large Language Models Hallucinate Electric Fata Morganas?
Abstract:AI hallucinations - that is, outputs which are made up, cannot be verified, or contradict the source material - are generally regarded as an engineering flaw to be dealt with. This paper contends that they also have philosophical significance when it comes to the question of machine consciousness. We examine the known causes of hallucinations in large language models - such as source-target divergence, discrepancies between training and inference, and overfitting - and we present two empirical investigations. In the first, we apply successive generations of the GPT model to ambiguous factual questions under different temperature settings, finding that higher temperatures result in plausible but incorrect answers while lower temperatures lead to factually accurate ones. The sampling parameters that cause a model to seem creative or spontaneous and thus more likely to pass behavioral tests of intelligence are the same ones that increase its hallucination rate. In the second, we look at an encoder-only model that has been trained on encyclopedic data and which answers questions of the same type factually and without embellishment, indicating that hallucinations are due to exposure to subjective and socially diverse training data rather than to the development of any cognitive ability. Using references to Turing, Searle's Chinese Room, the frame problem, and the cybernetic tradition of Wiener and Ashby, we claim that a model's self-reports of emotion or sentience come within the definition of hallucination, and that any future occurrence of machine consciousness might remain epistemically inaccessible since it would be indistinguishable from a sufficiently advanced hallucination.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2608.18816 [cs.CL]
(or arXiv:2608.18816v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2608.18816 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
Journal of Consciousness Studies 32 (11): 96-120. 2025
https://doi.org/10.53765/20512201.32.11.096 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 377 words
Related coverage
[AINews] not much happened today
Latent Space · Sep 10, 2026
Fact Grounded Attention: Eliminating Hallucination in Large Language Models Through Attention Level Knowledge Integration
arXiv · Sep 22, 2026
The death of humanity
Fortune · Oct 16, 2026
AI Futures: Racing to Lose, A sermon
LessWrong · Sep 28, 2026