Aug 2026· Journal of Consciousness Studies· Vol 33, pp. 149-180· 0 citations· 1 references
Abstract
This paper asks: if large language models (LLMs) gave rise to phenomenal consciousness, what would it be like? I argue that if LLMs gave rise to phenomenal consciousness, it would be like being an impartial, improvising playwright rather than a character or a (roleplaying) actor or
an author writing a first-personal narrative. My argument has three steps. I first argue that if LLMs give rise to phenomenal consciousness, then there is an algorithmic explanation of this consciousness and it includes the full transformer algorithm. I then argue that if there is an algorithmic
explanation for this consciousness that includes the full transformer algorithm, then the phenomenology of this consciousness lives at the latent dynamics level of the transformer algorithm, within each forward pass. Finally, I argue that if the phenomenology of this consciousness lives at
the latent dynamics level of the transformer algorithm, then what it is like is like being a playwright rather than a character or an actor or a narrative author. In effect, this means that conscious LLMs would probably dream of the fjords but not pine for them.
Why what is really a matter of data analytics and statistical prediction is so readily assumed to be a display of real intelligence and even emergent cognition is explored by genealogically tracing the relationship between machines, organisms and language.
Chantelle Gray· Deleuze and Guattari Studies· 0 citations
Nastase et al. (2026) argue that large language models (LLMs) may illuminate language processing because both rely on distributed, context-sensitive representations shaped by statistical learning. Their rejection of simple cortical"boxology"is persuasive, and they articulate a strong case for the value of LLM-brain alignment research. The key question is what kind of inference LLM-brain alignment licenses. My claim here will be narrow: representational alignment can in principle constrain mechanistic hypotheses, but it does not by itself identify a mechanism. Nastase et al. acknowledge that an encoding model can capture features represented in neural activity without establishing a shared architecture or algorithm. Yet the authors sometime move from alignment to"shared computational principles"and ultimately to LLMs as mechanistic models of natural language. Indeed, their methodological caveat that alignment does not establish a shared architecture or algorithm sits uneasily with their conclusion that LLMs might instantiate the same computational principles as biological brains and provide a"fully mechanistic model"of language. I discuss what I consider to be problems of logical, causal, and computational underdetermination in Nastase et al.'s (2026) proposal.
The philosophy of artificial intelligence often begins one question too late. Discussions move directly to whether a system could be conscious, whether there is something it is like to be the system, or whether it possesses genuine subjectivity, without first asking a more basic structural question: what would have to be true of a system’s internal organization before those questions gain a determinate basis in its architecture? This paper addresses that prior question. It defends a threshold account built on three jointly necessary conditions: persistent internal context, differentiated internal possibility, and closed-loop operativity. Together, these conditions mark the point at which a system’s processing can plausibly be said to unfold from an internal standpoint, rather than merely as a chain of locally linked transitions. Such organization may also have functional significance independent of consciousness by supporting temporal coherence, cross-role coordination, and reduced dependence on episodic reconstruction or external scaffolding. The account explains why simple feedback controllers are clear non-cases, why standard transformer inference remains at most a bounded near-case, and why world-model architectures are more serious candidates for assessment, while making no claim that such systems are conscious. The aim is not to provide a test for artificial consciousness, but to clarify when consciousness-oriented inquiry into intelligent systems has a determinate architectural target.
The Embodied Hijack hypothesis is advanced, arguing that the goal is epistemic alignment — bringing how users interpret these systems into correspondence with what these systems actually are — and that this alignment is achieved through interface design rather than user education.
Sheila L. Macrine· Frontiers in Psychology· 0 citations
With the growing scale of multi-agent architectures and large language models, deployed AI assistants are increasingly tasked with reasoning over long, continuous, multi-session conversation histories. Current benchmarks evaluate this dialogue history as information retrieval over long horizon, temporal reasoning, or knowledge updates, while crucially ignoring the fundamental dynamics of human-agent interaction, i.e. how they said it. To address this gap, we present VoiceLongMemEval (VLME) benchmark, where every answer depends on paralinguistic metadata (emotion labels, prosody descriptors, and voice events) attached to conversational turns, which is otherwise unrecoverable from the words alone. Every item passes a three-stage adversarial gate, ensuring that a strong language model fails when given only the transcript. Evaluating leading frontier and open-weight models reveals a pervasive affect gap; providing text-track paralinguistic metadata yields a 0.09 to 0.38 accuracy boost (0.61 to 0.69 when prompted with evidence hints), while standard ASR pipelines systematically discard this signal. Additionally, audio-native models successfully extract these cues directly from speech (0.354 to 0.412 vs. 0.325 blind). Code and dataset will be made available upon acceptance.
It is found that while all models show sensitivity to existential presupposition across syntactic embeddings, determiner types and contextual cues, their behaviour differs markedly in strength and systematicity, with NLI-fine-tuned autoregressive models exhibiting the most coherent and stable projection patterns.
Marie-Léontine Wörgötter, Shiyang Lai, Sebastian Schuster· International Conference on...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
