At the outset, when considering the possibility of “artificial consciousness,” it is essential to be clear about what aspects of consciousness we are considering. We agree with Seth’s statement (Section 1.1) – based upon Nagel (Reference Nagel1974) – that “for a conscious organism, there is something it is like to be that organism” and that if it “feels like” something to be a conscious system – there is a conscious experience happening – whereas it doesn’t feel like anything to be an unconscious system – there is no conscious experience happening.” Note that Nagel’s emphasis is clearly on the “feeling” aspect of consciousness. Seth prefers the term “consciousness” over “sentience,” because he suggests the latter is used inconsistently in the literature and is prone to confusion. “Sentience,” however, has been used widely and consistently in animal-consciousness studies for the past decade (Birch, Reference Birch2024; Chittka, Reference Chittka2022; Harnad, Reference Harnad2016). Thus, we prefer “sentience” over the broader term “consciousness” because its Latin root means “to feel,” so it emphasizes consciousness as feeling, not just as aspects of cognition or computing. So, in our view, it is the “feeling” aspect of consciousness that is the key issue in the possibility of “conscious AI.”

Second, Seth’s argues – correctly in our view – that the neural processes behind consciousness are tied to a living nervous system and therefore are not substrate-independent as the Turing-computational functionalism behind standard AI demands – thereby indicating such AI cannot attain consciousness. Indeed, Seth provides a comprehensive list of the differences between standard silicon-based computer circuits and living neurons/brains, and these would most likely preclude consciousness in conventional AI systems (Section 3). These differences include the following: brains are embodied but Turing computers are disembodied; recurrent processing versus linear processing; extreme integration of information versus lesser integration (the generative entrenchment concept in Section 3.3); spike-wave synchronization is used by neurons alone; and the conscious brains alone sense the external environment through multiple sensory modalities for survival and propagation (Section 4.5). Seth identified what he argues is the “real problem of consciousness” and characterized it as a truly scientific challenge: “as we become better able to explain, predict, and control properties of consciousness by underlying mechanisms, the sense of mystery … will dissolve” as did the mystery of life in the past two centuries (Section 4.4).

Third, Seth mentions hierarchical emergence (Section 3.5) in the context of “substrate flexibility” in physics. Here, we expand upon this point with reference to consciousness because we – and many others – have argued that it is essential to view sentience – and more broadly consciousness – as an emergent process of brain functions (Broad, Reference Broad1925; Chalmers, Reference Chalmers, Clayton and Davies2006; Feinberg, Reference Feinberg2024; Reference Feinberg2025; Feinberg & Mallatt, Reference Feinberg and Mallatt2020, Reference Feinberg and Mallatt2016; Kim, Reference Kim1998, Reference Kim, Clayton and Davies2006; Searle, Reference Searle1992; Thompson, Reference Thompson2007)

Emergence is a property of hierarchical systems in which lower levels interact with higher levels. In general terms, an emergent feature is a novel, aggregate feature of the system that is not present in the parts alone. Commonly cited biological examples of emergent processes are respiration, metabolism, and digestion as they function in the body of vertebrate animals (Salthe, Reference Salthe1985; Searle, Reference Searle1992). Emergent properties are often described as “more than the sum of their parts” because the parts interact at multiple hierarchical levels and those interactions are the key to the emergence of novel system features (Ahl & Allen, Reference Ahl and Allen1996; Bedau & Humphreys, Reference Bedau and Humphreys2008; Clayton, Reference Clayton, Clayton and Davies2006; Vintiadi, Reference Vintiadis2013). In the nervous system of vertebrates, for instance, consciousness emerged as a novel system feature when new higher levels evolved through the expansion of the forebrain. Therefore, like life, sentience is best understood as an emergent feature of a complex system, in this case, the living central nervous system.

An additional factor that links emergence, systems theory, and sentience is the feature of complexity. There is substantial theoretical support for the role that complexity plays in increasing emergent-system properties (Ladyman & Wiesner, Reference Ladyman and Wiesner2020; Mitchell, Reference Mitchell2009; Vintiadis, Reference Vintiadis2013), and the most complex systems in nature are all biological, from cells to multicellular organisms to animals with complex brains. Further, when we trace the evolution and neurobiology of sentience, there is a close correspondence between the neurobiological complexity of the nervous systems and the emergence of sentience. This finding is based upon the view that organisms that are sentient are those whose brains are documented to create higher-order mapped neural sensory representations of the world, possess the neural infrastructures that enable interoceptive-affective feelings with positive versus negative valence, and the capacity to hierarchically combine all these diverse neural processes into an integrated sentient experience. These conditions are only met by the complex-brained vertebrates, arthropods, and cephalopods like octopuses (Feinberg Reference Feinberg2024, Reference Feinberg2025; Feinberg, & Mallatt, Reference Feinberg and Mallatt2020, Reference Feinberg and Mallatt2016; Mallatt et al., Reference Mallatt, Blatt, Draguhn, Robinson and Taiz2021; Mallatt & Feinberg, Reference Mallatt and Feinberg2021).

These considerations indicate that the living biological substrate and the neural parts of sentient brains – via the constraints imposed upon them by their biological, adaptive, and evolutionary nature – are essential to the emergent properties of sentience, an example of what Seth calls the strong substrate-dependent hypothesis (Section 5.5). Consistent with the principles of emergence, sentience and more broadly consciousness depend upon the physical structure and processes of the parts in addition to the neuron-based, multi-leveled hierarchical interactions of these parts. Therefore, it is highly emergently substrate-dependent – in contrast to the processes of the conventional, substrate-independent, Turing-based AI systems.

Finally, today’s computers – despite their advanced AI capabilities – are not capable of “feeling,” and thus are neither sentient nor conscious. Seth hypothesizes that if someday artificially intelligent entities may be made more life-like, it is at least conceivable that they could have “conscious AI.” While this possibility is something that cannot be definitively predicted, we feel it is quite likely that whatever “sentient-like” features (if any) these non-living, non-biological entities display, they will not be “experienced” in the same way that animal sentience is personally experienced. This is because computer non-biological substrates and their organizational properties will likely be so vastly different from living, biological, and neurobiological consciousness that is the result of billions of years of evolution, as sentience and consciousness gradually emerged from life.