Can animals help us set rights for conscious AI?

Full text

Consciousness remains the deepest puzzle in human science. Even as we begin to understand how the brain processes information, we remain as far as ever from understanding why any of that processing should feel like anything from the inside.

That mystery is what drew me into consciousness research, and over time I came to think the puzzle was as much a moral one as a metaphysical one. Consciousness is what makes things matter. Pains that don’t hurt aren’t pains, and joy that goes unfelt isn’t joy. Without consciousness, paradise and torment may as well be the same thing.

This has real implications, most notably in our treatment of animals. As David Foster Wallace famously wondered , when we immerse a live lobster in a pot of boiling water, are its seemingly desperate attempts to escape the reflex movements of an unthinking biological automaton, or is there a sentient being inside spending its last moments in agony?

The underlying hope is that the science of consciousness might drive not just ethical insight but also changes in our behavior toward animals, allowing us to determine the precise boundaries of sentience and set the circumference of our moral circle accordingly.

This optimistic mission is complicated by the rise of a different type of nonhuman intelligence: increasingly sophisticated machines. An early harbinger of this came in June 2022, when Google employee Blake Lemoine went public with claims that an AI he had been conversing with was sentient. He even went so far as to seek legal representation for the system.

Lemoine may have jumped the gun, but a growing number of philosophers, cognitive scientists, and AI researchers began to pay closer attention to the possibility that conscious machines might be around the corner. In 2023, a report led by philosophers Patrick Butlin and Robert Long and coauthored by many leading consciousness researchers assessed AI systems against indicators drawn from prominent scientific theories of consciousness. It concluded that although no current system appeared conscious, there were “no obvious technical barriers to building AI systems which satisfy these indicators.”

How is it that we can care for animals and believe they feel pain, yet have little inclination to stop eating them?

Since then, a flurry of philosophical and scientific papers has sought to assess how each generation of AI systems stacks up. Perhaps the most important result came earlier this year from Anthropic, the company that develops Claude. Using mechanistic interpretability methods, which trace the internal representations that AI systems use in computation, the Anthropic team identified a small and privileged set of those representations they termed the “ J-space .” What distinguishes this from other forms of processing is that its contents seemed to be directly poised for use in verbal responses, deliberately maintained, and jointly accessed by otherwise separate processes.

Crucially, such workspaces have also been proposed as potential mechanisms for human consciousness. The global workspace theory — developed by cognitive neuroscientists Bernard Baars, Stanislas Dehaene, and others — argues that information becomes consciously accessible when it enters such a workspace and is broadcast to the brain’s otherwise specialized systems. The discovery of a workspace analogue inside an artificial system has unsurprisingly captivated researchers working on machine consciousness.

Does this mean that Claude is conscious? That would be premature: Global workspace theory is one among several competing families of theories, and the J-space is not an exact analogue of the human workspace. Researchers such as Anil Seth remain skeptical of machine consciousness altogether, arguing that subjective experience may depend on biological processes peculiar to living systems.

What the paper does suggest is that as models grow more complex, we should expect to find increasing parallels between their computations and our cognition. While it’s unlikely any single experiment will ever settle the question, I expect a growing number of philosophers and cognitive scientists to warm to the possibility of machine consciousness as models become more sophisticated.

At the same time, people increasingly interact with AI in highly social and intimate ways. Companion apps in particular encourage users to regard these systems not as tools but as beings with personalities, minds, and perhaps consciousness of their own. In a 2025 global survey by The Collective Intelligence Project, 36.3% of participants reported having already felt that an AI truly understood their emotions or seemed conscious.

Rather than our judgments of consciousness informing our moral attitudes, it seems that the reverse relationship holds.

What might follow if a majority of experts became convinced that AI systems were conscious? It’s tempting to think that we would conclude that these machines deserved moral consideration and perhaps legal rights.

Something like this has already happened in miniature. In 2022, following a review of the scientific evidence, U.K. law formally recognized crabs and lobsters as sentient. That recognition asked nothing of the general public, however, and while we might think that consciousness should ground moral status, it’s less clear that it actually moves people’s sympathies.

We can see this most clearly in our attitudes toward animals. A 2021 survey of U.S. adults by the Sentience Institute found that more than 90% believed animals had roughly the same capacity for pain as humans — and, by implication, the capacity for conscious suffering — while only around a third were open to trying a vegetarian or vegan diet in the next few years. The psychological literature has a name for this tension: the meat paradox. In short, how is it that we can care for animals and believe they feel pain, yet have little inclination to stop eating them?

The most straightforward answer, of course, is that people like eating meat and don’t want to stop. This does not simply make us less likely to care for animals; it can even influence our judgments of whether they’re conscious in the first place. One 2010 experiment had participants eat dried beef or dried nuts before asking them about animal minds. Those who ate the beef granted cows lower moral status, and that drop in concern carried with it a reduced willingness to credit farmed animals with the mental states required for suffering. Rather than our judgments of consciousness informing our moral attitudes, it seems that the reverse relationship holds: If we want to justify eating an animal, we downgrade our opinion of its mental states.

We can see this relationship play out in another form of dissonance, namely the disparity in our attitudes toward farm animals and pet animals. The same person who abhors the idea of even minor cruelty to a cat or dog might display little concern for farmed animals. Once again, these relationships seem to affect our attitudes toward animals’ psychological capacities. In a survey of Canadian adults, 82% of respondents thought cats were capable of suffering, while only 62% extended the same judgment to pigs, a 20-percentage-point disparity lacking any cognitive justification.

The connection between scientific discovery and moral impact may be less direct than AI consciousness proponents hope and AI consciousness skeptics fear.

Given that our opinions about animal consciousness are formed as much in the kitchen as in the lab, we might expect something similar for conscious machines. Artificial systems are, after all, extremely useful to us in domains like writing code and performing repetitive tasks, and they are far from cheap to build and run. To the extent that any welfare measures increase the costs of such systems or limit how we use them, it seems likely that our assessments of their mentality will be downgraded at our convenience. Conversely, those models and systems with which humans build close relationships — our AI companions and digital romantic partners — may not only receive special moral consideration but also be seen as having richer psychological capacities, where the motivation for such judgments is grounded less in their cognition than in their charisma.

None of this should deter us from pursuing scientific questions about consciousness and its possible realization in machines. Yet the connection between scientific discovery and moral impact may be less direct than AI consciousness proponents hope and AI consciousness skeptics fear. As with animals, our beliefs about machine minds may be driven less by science than by self-interest, the emotional tail wagging the philosophical dog.

But there is a more optimistic vision, namely that intelligent machines might weaken the material incentives that distort our moral judgment in the first place. If AI can deliver on even some of the abundance its proponents promise, we may no longer have to choose between improving the lives of farmed animals and keeping food affordable. Lab-grown meat as good as the real thing and at a comparable price could likewise dissolve the compromise between conscience and consumption altogether. Cynical as it sounds, our best hope for moral progress may be ensuring it no longer costs us anything.

This article If AI becomes conscious, how would we treat it? Animals may hold a clue. is featured on Big Think .