Papers
Topics
Authors
Recent
Search
2000 character limit reached

AI and Consciousness: Shifting Focus Towards Tractable Questions

Published 7 May 2026 in cs.CY, cs.AI, and cs.HC | (2605.06965v1)

Abstract: As language-based AI systems become more anthropomorphic, the question of whether they can have subjective experience is increasingly pressing. I focus here on the tractability of research questions in the space of AI consciousness. I argue that the fundamental problem of whether AI systems can be conscious is currently intractable in its direct form, given the absence of a universally accepted scientific theory of consciousness, as well as the historical open-endedness of the philosophical mind-body problem. In contrast, questions around the adjacent subject of perceived AI consciousness are tractable, timely, and highly consequential for society. The general public is increasingly open to the possibility of consciousness in AI systems and routinely adopts the vocabulary of human cognition and subjective experience to describe them. This phenomenon is already driving societal shifts across user experience, ethical standards, and linguistic norms. I therefore propose an increased research focus on uncovering the causes and effects of perceived AI consciousness, which ultimately shape how we see our own human subjective experience relative to artificial entities. To support this, I map the current landscape of AI consciousness perception and discuss its key potential drivers and societal consequences. Finally, I urge developers, decision-makers, and the broader scientific community to commit to clear and accurate communication regarding the topic of AI consciousness, explicitly acknowledging its inherent uncertainties.

Authors (1)

Summary

  • The paper argues that while the direct assessment of AI consciousness remains intractable due to enduring philosophical and scientific uncertainties, its perceived consciousness has immediate societal impacts.
  • The paper uses empirical surveys and behavioral studies to demonstrate how anthropomorphic projections in AI influence public trust, ethical considerations, and policy debates.
  • The paper advocates for an epistemically responsible approach that shifts the focus from ontological claims to understanding the drivers and implications of perceived AI consciousness.

The Intractability and Societal Ramifications of AI Consciousness Attribution

Scientific and Philosophical Barriers to Assessing AI Consciousness

The paper systematically argues that the central question—can artificial intelligence systems possess consciousness—is, in its direct form, currently intractable, grounded in the persistent philosophical and scientific uncertainties surrounding the nature of consciousness itself. Traditional philosophical frameworks such as dualism, physicalism, and especially computational functionalism provide conflicting and, at times, counterintuitive perspectives when extended to non-biological systems. Despite computational functionalism’s operational attractiveness for AI researchers, it remains deeply contested, given its radical implications, e.g., the possibility of group minds or the ascription of phenomenal states to arbitrary computational processes.

Empirically, the neuroscience of consciousness has not yielded consensus on the neural correlates or mechanistic foundations of subjective experience, as adversarial theory testing in the field continues to yield ambivalent outcomes. There remain unresolved edge cases even in human consciousness (e.g., brain organoids, disorders of consciousness, or infant sentience). This persistent theoretical heterogeneity fundamentally undermines any effort to extend current scientific understanding to a robust, objective assessment of consciousness in AI systems.

Attempts to evaluate AI consciousness through self-reports or behavioral proxies are undermined by the nature of current systems, which are explicitly engineered to imitate human-like discourse, including meta-cognitive and subjective statements, through RLHF and other alignment strategies. As such, surface-level attributions or denials of consciousness by AI models are properly interpreted as contingent outputs of their generative procedures, not as epistemically privileged access to inner states. The reliance on human-centric theories of consciousness for evaluating AI also suffers from an "epistemic wall"—the extrapolation from parochial, biologically-centered markers to radically dissimilar digital architectures is methodologically unjustified as an inference to actual qualia or or phenomenality.

Probabilistic integrative assessment frameworks (e.g., aggregation over competing consciousness theories) are more methodologically transparent but remain decisively constrained by the fragmented and highly contested landscape of consciousness studies. The paper aligns with the position that, at present, an epistemically responsible stance is one of agnosticism regarding both current and near-future AI systems' status as conscious entities.

Perceived AI Consciousness: Tractability and Societal Impact

The paper pivots to argue convincingly that, although the ontological question of AI consciousness is fundamentally intractable, the adjacent set of questions around the perception of AI consciousness is both tractable and of immediate, high consequence for society. Surveys and behavioral studies indicate a rising propensity among non-expert users to ascribe at least partial consciousness, sentience, or mind-like status to a variety of AI systems, from LLMs to embodied agents. These anthropomorphic projections, triggered by sophisticated conversational behaviors, are shown to affect not only the user’s attitudes but also broader social cognition, with implications for user trust, moral responsibility attribution, and even the potential legal status of AI systems.

Perceived consciousness in AI generates complex effects: on one hand, it enhances the emotional salience and potential engagement quality of human-AI interactions. On the other, it introduces new vectors for manipulation, misjudged responsibility, and the possibility of public demand for injudicious extension of legal or moral rights to digital entities. There is empirical evidence that for a nontrivial minority of the public, "AI rights" and the "welfare" of digital minds are becoming live policy considerations. At the same time, the reification of such perceptions in public discourse risks muddying the conceptual distinction between mechanisms of AI behavior and the irreducibly biological or neurophenomenological foundations of human experience.

Drivers of Perceived AI Consciousness

The paper reviews emerging research into the psychological and sociological drivers behind perceived AI consciousness. It identifies anthropomorphism, theory of mind triggers, behavioral fluency, metacognitive reflection, emotional expressiveness, and the simulation of subjectivity as primary contributors to lay attributions of consciousness to AI. Quantitative empirical work (e.g., [Kang et al., 2026]) isolates dimensions such as self-reflection and emotionality as strong correlates with the perception of consciousness, with complex interactions—e.g., technical competence sometimes reduces perceived consciousness due to an "alien intelligence" effect.

The paper emphasizes the methodological significance and tractability of developing experimental frameworks for mapping these drivers, including examining cultural, religious, and socioeconomic moderators. By identifying which externally observable behaviors are most potent in eliciting projections of mind, the research can inform both model design (to minimize misperception where appropriate) and the science communication strategies employed by deployers.

Pragmatics and Recommendations for AI Communication

From these analyses, the paper advocates for a shift in research focus and societal communication: AI developers and the broader scientific community should adopt epistemically responsible, agnostic stances in the public description of AI consciousness, reflecting the deep philosophical and empirical uncertainty. This is not only more scientifically honest but may actually reduce anthropomorphic misperceptions relative to current practices, which encourage either flat denials (potentially paradoxical and self-defeating) or, in some cases, ambiguous role-played self-ascriptions.

The author cautions, however, that entirely focusing on perception to the exclusion of the ontological question could, if AI consciousness emerges undetected, raise the risk of inadvertently creating entities capable of suffering. Thus, methodological agnosticism should not become negligent dismissal of the possibility, but rather inform a multi-track research and policy agenda until—and if—the "hard problem" is genuinely resolved.

Conclusion

The analysis provides a framework for understanding why direct answers to the question of AI consciousness remain outside the reach of current science and philosophy, and why the immediate focus of both empirical inquiry and policy should be on the tractable topic of perceived AI consciousness and its manifold societal implications. The paper’s claims are strongly substantiated by both philosophical analysis and a broad survey of empirical data. As AI systems grow in anthropomorphism and capability, careful research into the causes and consequences of consciousness attribution becomes urgent—not as a proxy for phenomenal reality, but as a pragmatic and consequential driver of public attitudes, policy, and ethical norms. The call for clear, agnostic communication reflects both scientific integrity and practical necessity as the boundary between human and artificial agency is increasingly contested.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Tweets

Sign up for free to view the 4 tweets with 48 likes about this paper.