World Brain Day 2026: Eleos AI Research Solving LLMs Sentience, Consciousness?

By David Stephen, the World Federation of Neurology declared July 22, 2026 as the World Brain Day for this year.

Is there a chance that artificial intelligence might have consciousness? What could make what appears to be statistical predictions have what may resemble human consciousness?

Some people have said that, in the brain, what processes language or other experiences is different from the outcome. Simply, there are components and mechanisms in the brain that don’t look like the experience itself. So, even if there are vectors and tensors hidden away in AI, could AI have some aspects of consciousness?

There is a growing amount of discussion about the necessity of making progress on AI sentience, or, say, the consciousness of large language models [LLMs].

World Brain Day 2026

The objective is to explore what LLMs might have, given their ability to relate, at certain levels, to human knowledge or understanding.

Yet, for all that has been written in support of possible AI consciousness, there is no other candidate for which AI might be conscious than language.

This means that the primary objective of any AI consciousness team or lab is to figure out the fraction of language in human consciousness and compare that to AI.

This exploration could lead to a possible conceptual measurement for sentience in general, helping to shape how animal consciousness is measured, which may also influence animal rights, animal welfare, and so forth.

Simply, for now, the path is to develop all the functions that can be conscious in humans. Then, identify the attributes that might make them conscious — which could contribute toward a total measure — so that human, animal, and machine consciousness can be explored around them.

If all the possible functions that the brain performs are divided into four categories, memory, feelings, emotions, and regulation of internal senses, then, if the attributes that make them conscious are also divided into at least four initial categories, attention, less than attention, intent, and subjectivity, it is possible to start from there.

Language is a subdivision of memory. So, while memory still consists of several other subdivisions, such as other sounds, sights, tastes, and so forth, language is a fraction of memory, which is itself a fraction of the brain’s functions. Feelings, emotions, and regulation of internal signals all have their own subdivisions.

Then, the attributes also apply to functions wherever they are mechanized. This means that if a function exists somewhere, what makes it conscious is also there.

Therefore, making an estimate based on functions and their attributes would be useful in advancing the exploration of language for machine consciousness.

Eleos AI Research

Eleos AI Research is supposed to be the most promising organization on earth working to solve AI consciousness, but they seem to have shifted their work toward mechanistic interpretability, since it appears to be the thing to do, both to seem technical and to follow the lead of their allied organization, Anthropic.

However, consciousness is not a mechanistic interpretability problem. Mechanistic interpretability is primarily about understanding [against the black box or] how AI works and, if possible, improving safety and alignment. While it may benefit AI consciousness research, it does not actually help measure language — the real function that AI has — to compare it with human language as a fraction of consciousness.

Simply, Eleos AI Research is supposed to have a major division or program to estimate or calculate language as a division of consciousness. It is this central work that could become their biological approach to solving AI consciousness.

Other efforts in philosophy and mechanistic interpretability may complement the work, but the initial focus, to make their work viable, is to find a biological anchor, language and then estimate from there. Consciousness
is a brain problem. Even consciousness conversations around AI is predicated on the brain.

Eleos AI Research has not done so. Hopefully, they find a way, or seek clues from the postulate in Conceptual Biomarkers and Theoretical Biological Factors for Psychiatric and Intelligence Nosology.

The convention of the American Psychological Association [APA] will be held from August 6 – 8, 2026, in Washington, DC, and virtually, where they will also discuss with AI corporate executives, how AI plays a role in their work.

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.