By David Stephen
In the human brain, there is no difference between intelligence and consciousness. Or simply, the human brain has shown that if intelligence is somewhere, consciousness is — or could be — there.
In the brain, what has been proven [by neuroscience] to mechanize functions are neurons and their electrical and chemical signals. So, whether the mechanism is for consciousness or for intelligence, it is the electrical and chemical signals — of neurons — that are responsible.
In the brain as well, what processes language are the electrical and chemical signals. This means that the configuration for what it means to have memory, feelings, emotions or regulations of internal signals are held by electrical and chemical signals, not neurons.
Why, because neurons are cells and have a near fixed anatomy, without the ability to adjust widely, to configure or represent all functions. So, the options are electrical and chemical signals, conceptually.
Now, if large language models [LLMs], with their intense optimization algorithms are assumed to be intelligent, could that mean that they might be conscious, given that consciousness and intelligence are parallel in the human brain?
Also, if the proper use of language, for humans, is a conscious process, could that mean that a proper use of language by LLMs is a conscious process?
These are the basic questions for any organization or team that is seeking to answer the AI sentience conjecture. The brain, the human brain, the brain. Nothing more. Nothing less.
Even as the topic of AI consciousness has gone mainstream, the two major organizations at the center of it, Eleos AI Research and Anthropic have research and efforts that have proved fruitless for more than a year.
Anthropic is a successful LLMs company but a failed company in AI consciousness research and a failed company in figuring out the biology of human intelligence, in spite of Claude being great at biology and the CEO being a biological scientist as well.
Anthropic in AI consciousness is feeling like Uber autonomous driving rush circa 2014. They didn’t have what it took. It did not matter everything they were and threw at it, they failed. Anthropic is already on the same missed trajectory, in biological-to-machine consciousness and intelligence.
What is worst with Anthropic is that even with their state-of-the-art mechanistic interpretability research that should inform a draw into machine mind and human mind, has not made them provide any particular riveting answer.
What is human intelligence in the brain? How is it parallel to artificial intelligence? This is not just a question for the consciousness problem but to also track the pathway to artificial general intelligence [AGI].
Where is human intelligence in the brain? Some studies point to the prefrontal cortex. OK, what goes on there that is different from some other part, say the hypothalamus, that makes intelligence domicile in one and not the other?
In both centers, there are clusters of neurons. These clusters can be assumed to contain sets of electrical and chemical signals. So, a postulate could be that there are near-dedicated configurations, assembly or formations of sets of electrical and chemical signals, that specify what happens in different circuits.
Simply, there should be a model of what intelligence could be in the brain and then what consciousness could be. Then the definition of a standard, where machines can be considered.
For example, if functions can be categorized as memory, emotion, feeling, and regulation of internal senses. Then if the factors [or attributes, or measures of the extents] of functions are attention, pre-attention, subjectivity, and intent. Then it is possible to find how AI compares.
Simply, all functions are mechanized by the interactions of electrical and chemical signals. However, the extents to which they interact is called their attributes or factors. Now, since AI does not have electrical and chemical signals, it is possible to compare with the divisions of functions [memory, emotion, feeling, regulation of internal senses] and their factors [attention, pre-attention, subjectivity, intent].
Now, what AI has is intelligence and language. Both intelligence and language are subdivisions of the function, memory. AI does not have emotions, feelings or regulation of internal signals.
So, language and intelligence, then what factors [attention, pre-attention, subjectivity and intent] apply and can be used to estimate a fraction of LLMs sentience?
This is how to make progress in the machine consciousness question, where there is a direct peg to the electrical and chemical signals.
Philosophers as Labelists
Eleos AI Research and Anthropic are seeking to solve consciousness from a philosophical angle. But really, how much of a philosophical problem is human or machine consciousness? Human consciousness is completely concerned with neurons — and electrical and chemical signals. AI consciousness is concerned with memory [language and intelligence]. Language and intelligence can be linked with human language and intelligence, as mechanized by electrical and chemical signals.
So, if the objective is electrical and chemical signals, which is the forte of conceptual brain science, or theoretical neuroscience at least, what is a philosopher to do?
There have been several papers written about AI consciousness in the last 18 months, by philosophers, all they have done is labels — and more labels and just labels. Nothing about a standard, no delineation of functions or the factors of functions, no reference to electrical and chemical signals. Like nothing close.
Philosophers are labelists in the consciousness problem. Hard problem does not mention what components of the brain to focus or what preliminary mechanism would be necessary. The same can be said for any philosophical label, no matter how cool it sounds.
Also, intelligence in the brain, is much harder to solve than consciousness. Intelligence, at least in the era of AI, is more important too. What exactly is human intelligence in the brain? What are they types, their mechanisms and components? Assuming a company answers this, and then presents a display, it can be useful for learning, memory, problem-solving modeling, and so forth, it would be a major advance and intense revenue magnet.
This however has not been the case. Eleos AI Research and Anthropic are trying to solve AI consciousness with philosophical labels. It won’t go anywhere, ever. There can be new labels or assumptions, but the problem starts with the brain and ends with the brain, extending to how AI compares, with language and intelligence, as divisions of memory.
And human memory, assuming Anthropic had a model, is the channel to making artificial general intelligence. Humans have generalized memory which helps to results in generalized intelligence.
Simply, human memory does not store a unique configuration of everything. It does collective configurations. So, a fan, door, table and so on are collectives, so interpretations are done with the collective. Aside that, there are overlays of collective configurations with similarities as well, like the one of a fan to an air conditioner, or that of a door to a window. This allows for memory to do some primary sorting work that makes intelligence easier.
So, if there would be AGI, it is likely that the data for which training is made, would be stored differently from what is currently obtainable.
All of these can emanate from a properly executed AI consciousness research, something that Anthropic lacks, as well as Eleos AI Research and Google DeepMind — with a recent philosophical hire.
Anthropic is exploring to solve biology with Claude. The most important biology, with opportunity for explosive answers and revenue is human intelligence, not just drug development or more proteins.
There is a recent [June 19, 2026] report on Reuters, US scientist John Jumper to leave Google DeepMind for Anthropic, stating that, “Senior research scientist John Jumper said on Friday he would leave Google DeepMind to join AI startup Anthropic, the latest high-profile departure ?at the Big Tech giant’s AI lab.”
“Jumper’s surprise departure comes just days ?after Noam Shazeer, a vice president of engineering at Google and co-lead of its Gemini AI models, said ?he ?would leave the company to join IPO-bound OpenAI.”
“Anthropic is hosting a science event on June ?30. The ?startup did not immediately respond to a Reuters request for comment regarding Jumper’s new role.”
David Stephen does research in conceptual brain science and served as a visiting scholar in medical entomology at the University of Illinois at Urbana-Champaign. He did computer vision research at Rovira i Virgili University in Tarragona.
See more breaking stories here.
