Part of an occasional series: ideas from outside the usual solarpunk reading list.
Here’s a question that used to sound like science fiction and increasingly doesn’t: an insect, a factory-farmed animal, and an AI system might all, for different reasons, be beings whose inner experience — if they have any — we genuinely can’t verify. Do we owe any of them anything? And if the honest answer is „we don’t know if they’re sentient,” does that uncertainty let us off the hook, or does it obligate us to something anyway?
Jeff Sebo, a philosopher who directs NYU’s Center for Mind, Ethics, and Policy, spends a short, sharp 2025 book — The Moral Circle: Who Matters, What Matters, and Why — arguing for a specific, carefully limited answer: uncertainty doesn’t excuse us. It obligates us to something, just not to everything equally.
The problem with the two obvious answers
There are two easy positions here, and Sebo thinks both fail.
Demand proof first. Set moral consideration aside until we’re confident a being is sentient — that it can actually suffer, prefer, or experience anything at all. This sounds appropriately cautious, but it sets an almost impossibly high bar. We don’t have a reliable, agreed-upon test for sentience even in cases we’re fairly confident about (most people believe dogs are sentient; we still can’t prove it the way we can prove a chemical reaction). Demand proof first, and you risk a catastrophic exclusion error: treating a genuinely sentient being as if it weren’t, simply because certainty never arrived.
Extend full consideration to everything, just in case. The opposite move — grant equal moral weight to anything that might conceivably be sentient — collapses under its own weight almost immediately. It offers no way to prioritize among billions of insects, uncertain edge cases, and clear paradigm cases like a suffering child, and in practice becomes unusable as a guide to actual decisions.
Sebo’s alternative: weight by probability
Sebo’s proposed middle path is a precautionary, probabilistic approach. Instead of a binary in/out test, assign moral status in degrees, proportional to the probability that a being is sentient, agentic, or otherwise morally significant, based on the best available evidence. A being we’re fairly confident is sentient (a dog, a pig) gets substantial moral weight. A being where the evidence is weaker or more contested (an insect, a plausible near-future AI system) gets some weight — not zero, not equal to the clear cases, but real weight scaled to genuine uncertainty rather than to convenience.
This does real philosophical work. It avoids the „grant everything equal status” trap, because probability-weighting is not equal weighting. And it avoids the „wait for proof” trap, because it treats moral consideration as owed proportional to a reasonable evidential estimate, not withheld until certainty arrives — certainty that, for many of the beings in question, may never arrive at all.
Where the book actually goes
The Moral Circle runs three registers at once, and moves between them deliberately rather than staying in one lane. There’s philosophical argument — working through the competing theories of what actually grounds moral status: sentience, agency, rationality, and how they interact. There’s applied ethics — what, concretely, do we owe insects, factory-farmed animals, and AI systems under genuine uncertainty about each. And there are provocative thought experiments, pitched at a general reader rather than a specialist one: would you rather share a house with a roommate who turns out to be a Neanderthal, or one who turns out to be a robot — and what does your instinctive answer reveal about what you actually think grounds moral status? Would you save a few elephants or a much larger number of ants? A few biological animals, or a much larger number of digital minds, if digital minds turn out to be real moral patients at all?
The book is part of Norton’s „Shorts” series — deliberately compact, classroom-tested, written for a general audience rather than professional philosophers, which makes it an unusually accessible entry point into what’s normally a fairly technical debate.
The honest limitation
Sebo’s own framework has a real cost, and it’s worth naming rather than glossing over. Precautionary, probability-weighted moral status tells you that some non-zero weight is owed to an uncertain case. It does not tell you how much — the framework is better at establishing the principle than at settling the practical dial. One reviewer, Gellers, has flagged this as a genuine gap relative to approaches grounded in the specifics of an actual relationship (rather than in an abstract probability estimate) — the kind of relational ethics associated with writers like Donna Haraway, which locates obligation in the particular, lived bond between beings rather than in a calculated probability. That’s not necessarily a knockdown objection to Sebo’s approach, but it’s a real seam, and probably the most interesting place to keep pushing if you find the precautionary framework compelling but incomplete.
Why this matters beyond the thought experiments
Sebo’s project sits at the intersection of two conversations that don’t always talk to each other: the long-running animal-ethics tradition (going back at least to Peter Singer’s „expanding circle” argument) and the newer, faster-moving conversation about AI moral status. Treating them as one continuous question — rather than two unrelated debates that happen to both involve non-human minds — is itself a useful move, because the same underlying problem (how do we act under real uncertainty about another being’s inner life) shows up in both, and solutions developed for one plausibly transfer to the other. Whatever you conclude about digital minds specifically, the precautionary framework gives a genuinely usable tool for the broader question of how to treat uncertainty about moral status as an occasion for graduated care, rather than as an excuse to do nothing until an answer arrives that may never come.
Did we get something wrong?
This piece is drawn from our own reading notes, not a translation of the book. If something above misrepresents the argument, or just sounds off, tell us — publicly or privately, whichever you’d rather.
Want the fuller picture?
This is a survey, not the whole book — including more of the thought experiments and a closer look at the Gellers critique. The deep-dive version is here: deep dive PDF.
What should we read next?
This is one piece in a loose series drawn from our own research reading. If there’s a thinker, a book, or a tradition you think belongs in a series like this, we want to hear it.
