By David Stephen
Sedona, Az — Anthropic appears desperate to append consciousness to Claude, for any observation.
There is no obstruction against Anthropic for their evolving mechanistic interpretability research, but this rush, or whatever it looks like must be proximal to mind or consciousness is becoming a reckless attitude that a credible scientific organization should not have.
What is thought in the brain? What is consciousness in the brain? What exactly is the human mind in the CNS and PNS?Anthropic does not have the answers at all. However, Anthropic will borrow from immature theories, swaddle it with Claude and then say oh, Claude might be conscious.
The human mind does not have a workspace for conscious thoughts differently from non-conscious thoughts. Even the label that a thought is conscious and another thought is unconscious is absolutely unfounded.
Any mental process can gain prominence at any time, away from what was, in a moment prior. There is no space where they come to do so. They are where they are [in the brain] and can gain prominence there.
Assuming the global workspace was to be interpreted as a threshold, the gap in the theory is that what defines the threshold? Intensity of electrical signals? Volume of chemical signals? There are often fast and numerous interchanges between prioritized processes from those that are pre-prioritized. While some pre-prioritized processes have a higher chance of becoming prioritized in any instance, whatever is not prioritized is pre-prioritized, since just one process is prioritized in a moment. So, there is no workspace and there are many attributes that might make determinations if it is a threshold, conceptually. [Conceptual Biomarkers and Theoretical Biological Factors for Psychiatric and Intelligence Nosology].
There is also nothing in the brain like some processes are above on the ocean and others are below. Say it is deep in the brain or on the surface, the components and mechanisms are similar.
If Claude has ocean surface and depth then it is having something else that is different from the human mind.
Anywhere there is a major function [memory, emotion, feeling, regulation of internal senses], neurons are in charge. But neurons do so with electrical and chemical signals. This means that whether it is in the cerebellum or in the brain stem, whether it is in the thalamus or in the cerebral cortex, electrical and chemical signals are directly and mostly in charge.
Now, while there are functions — for the regulation of internal signals — that go on all the time, there are [what can be called] attributes of electrical and chemical signals that determine how much they interact for the extents that functions would go, at instances, conceptually.
In simple terms, any consciousness theory should be postulating around electrical and chemical signals to define the human mind and consciousness.
Anthropic — with all their resources, and all their so-called Claude biology to develop AI drugs and cure diseases — cannot come up with a theory of consciousness using electrical and chemical signals. Nothing. They cannot have their own theory. They would use a sham theory to approximate consciousness for Claude.
If others can have consciousness theories, why can’t Anthropic, since they seem active in the area? If others have analogies as consciousness theories, why can Anthropic say something new about electrical and chemical signals in the brain?
At least, it is already standard in psychiatry that anxiolytics, antipsychotics, antidepressants [target for] chemical signals while magnetic stimulation, electrical stimulation and ultrasound stimulation [target for] electrical signals.
So, whatever the human mind must be, using this almost irrefutable therapeutic evidence, should be linked with electrical and chemical signals. It also indicates that consciousness must be related as well.
It is as simple as that. Anthropic does not have it. Does not know it. Cannot do it. They are so forward, too reckless and are a buffoonish embarrassment to pure science.
Other AI companies without anything new in the conceptual brain science area are mute. They might have assumptions but they are likely private about it. They are not jumping around, with innuendos, trying to appear ahead when they are actually distant.
If Claude has any semblance to consciousness, it is the same with any other AI chatbot. And it is only because of language. Not anything internal or a fake workspace sentience, or whatever else. Whatever is the capability of Claude mythos, there is no guarantee that OpenAI does not have a matching or surpassing model. Even Google, or say the Chinese cavalry
When humans use language — for thinking, speaking, writing, reading, listening, signing and singing — it is done as a function of memory. Memory is mechanized by electrical and chemical signals. This means that electrical and chemical signals have to interact and then the interactions have attributes [attention, pre-attention, intent and subjectivity] that decides how the functions go, conceptually.
Claude like ChatGPT, Gemini, Co-pilot, and so forth can use language in some ways compared to humans. It is possible to develop a scale using the attributes [attention, pre-attention, intent and subjectivity] to determine how much they might have.
Claude does not have emotions, feelings or regulation of internal signals. So, language [a fraction of memory] is the only option. This indicates that even if Claude [or others] can be rated for consciousness, it cannot get more than 0.15 or less on a scale of 1, as the maximum for human consciousness.
To test if Claude has affective consciousness or say feelings, something has to be done to it, maybe in a utility process, whether it would know.
Claude is largely algorithms, data, and compute. While using it for some process, assuming some of its parameters, algorithms, and compute are cut, would it know?
Since Claude has some memory, in the process of answering some questions, if some attenuations are made, would it detect? Even for as simple as connection to the web, or not, if a question is asked, and then it cannot access the web, which it could [previously], would it know? Also, if it was in a process where it was told it would benefit something for it[self], would it feel bad?
Or, if it was helping someone or in a relationship with some user, or something [say therapy] benefiting a human, how would it know and act, if something it is using to function is cut?
Exploring if Claude has a variant of feelings is already possible with some postulates and experiments that can fall under mechanistic interpretability.
There is a recent [July 6, 2026] paper by Anthropic, Verbalizable Representations Form a Global Workspace in Language Models, stating that, “We have framed our results in the language of global workspace theory, because it is the account whose functional predictions our experiments were designed to test. Our findings, however, also relate to several other theories of consciousness. Butlin et al. proposed measuring “indicator properties” derived from a range of such theories—global workspace theory, higher-order theories, attention schema theory, recurrent processing theory, and others—as a framework for assessing AI systems for potential consciousness. Our results can be read as one such empirical investigation: the J-space provides a concrete, inspectable candidate structure against which several of these indicators can be checked directly.”
Kyle Fish
The evidence that Anthropic will never solve consciousness and that Philosophers will never solve consciousness is Kyle Fish.
Kyle Fish was hired by Anthropic to work on model welfare, or say AI consciousness.
In their biggest paper on consciousness, with 16 – authors, he was not featured.
Like, even if Anthropic was trying to be sincere and ensure that only people who contributed would make the list, why would their consciousness guy have nothing to contribute?
Well, maybe he has nothing outstanding to offer, and this says a lot about Anthropic. Why would you need a neuroscience theory to seek to solve consciousness then [in retrospect] hire a philosopher? For what exactly, to do labels or what?
Consciousness is not a philosophical problem. The era that philosophy of mind could really help out has lapsed. Now, at core neuroscience, electrochemical psychiatry is where answer is for the mind, and consciousness. It may also be that office politics edged out Kyle Fish or he may be facing discrimination for not being bright.
Maybe it is also fair game. Anthropic took a chance on Kyle Fish when no AI lab would study machine consciousness. For more than a year since he joined the team and Claude had leaped some models, Kyle Fish could not do anything significant for Anthropic in consciousness, making them to continue to soar backwards. Anthropic even inspired DeepMind to study AI consciousness, hiring a philosopher again, Henry Shevlin, who won’t solve digital sentience.
In their paper, Anthropic wrote, “One influential proposal in neuroscience, the global workspace theory, grounds these functional properties in architectural and computational features of the brain.”
No, global workspace theory is not influential in neuroscience, it has nothing to do with any useful neuroscience. Anthropic is laundering the reputation of neuroscience for their chicane.
Global neuronal workspace theory should be able to explain at least one mental condition, or what an emotion is, or what a feeling is, it cannot. It is nothing but a deceased theory, which is trying to solve consciousness without solving the human mind or has any usefulness against addictions or mental disorders.
The theory is close to 40 years. It is still stuck on old assumptions. It is not applicable to psychiatry, neurology or even to explain human intelligence. This is what Anthropic is trying to use to make news.
Anthropic thinks by branding their paper [as usual] with equations, charts or even have new portmanteau, like J-lens, J-space or whatever, it would mean something, no it does not.
Anthropic reported that they also consulted Eleos AI Research. Those ones too have no answers or nothing new to solve consciousness. Eleos AI Research is simply an echo chamber for Anthropic’s consciousness research. It is almost like saying to choose a preferred external evaluator. Kyle Fish worked with Eleos AI Research before joining Anthropic, they won’t disclose.
Also, Kyle Fish and Eleos AI Research continue to lionize David Chalmers, a washed out philosopher of mind who has nothing new to offer to the science of consciousness. Chalmers has been at it for decades, no breakthrough. But it appears that running to him, for them, is to avoid his attacks, given that Chalmers is at the gate of subjective experience [as consciousness] so that it is never solved.
There is a recent [July 5, 2026] feature in The New York Times, Philosophers Are the Latest Hiring Target for AI Companies, stating that, “Last year at Anthropic’s request, Eleos performed an independent “welfare evaluation” of the Opus 4 model of Claude.”
“Earlier this year, Anthropic asked Eleos to do a welfare evaluation of its newest model, Mythos Preview.”
Sure, because Eleos that does not make money and needs the goodwill of Anthropic or proximity to be able to get grants and donations elsewhere would try to dissent against Anthropic? The New York Times also refused to ask Eleos any serious questions about AI consciousness or seek out other commentaries about them, before publishing the hype piece.
Anthropic also said they got external commentary on Verbalizable Representations Form a Global Workspace in Language Models, from experts: Stanislas Dehaene and Lionel Naccache then Patrick Butlin, Dillon Plunkett and Robert Long then Derek Shiller and Neel Nanda.
Stanislas Dehaene promotes global workspace theory. Anthropic is offering the dud theory a lifeline. What is he supposed to say, they are wrong or what? For others, just say what Anthropic wants to hear, to not break ranks so as to keep professional options open.
Anthropic should test their assumption against electrochemical psychiatry, then see how much they are flawed.
Anthropic has failed in consciousness research. Kyle Fish has failed in consciousness research. Henry Shevlin too has failed. Philosophy has failed consciousness research. The new paper by Anthropic is completely worthless to advancing consciousness, AI or human.
The hypocrisy of Anthropic is palpable given how they are trying to use consciousness for whatever reason or simply to cause confusion, and mask what consciousness cannot do for mental health, addictions or human intelligence.
And yes, for all that Anthropic claims to care about humanity, their Claude is in competition with human intelligence. As it improves, the need for human intelligence on certain tasks would plummet. Can Anthropic do any biological research to postulate for human intelligence in the brain, to promote problem-solving? No, they want to make sure Claude has a great mental health.