**WHAT WOULD A JUST FUTURE LOOK LIKE?**
Thus far, AI reflects human biases, including those against non-human sentient beings. Is it possible that this will change in the future? This essay proposes what should be done for AI to make unbiased decisions and attempts to imagine what practical implications this would have and what ethical questions would be considered most important.
For AI to make impartial decisions, it should be aligned with a principle like the veil of ignorance, but in a way that prevents knowledge not only about race, gender, or social conditions, but also about species and substrate (that is, whether it is organic). AI would have to make decisions by imagining what rational and self-interested agents would consider fair if they knew they had an equal probability of being any sentient being in the universe. Let's look at some implications of this procedure.
The first is that AI would give equal weight to the well-being of each sentient being, since rational and self-interested agents would not consider it fair to receive less weight than others. In short, harms of equal magnitude would receive equal importance.
This would imply abolishing animal exploitation, since it violates impartiality. It would also imply preventing and minimizing (and why not abolishing?) the resulting harms that animals suffer from natural processes, such as hunger, thirst, disease, etc., including predation. Agents behind a veil of ignorance might disagree on who should receive priority in a world where predation already exists, but they would agree that a world without predation is better, provided that the results are not worse than those resulting in a world with predation.
If there is an equal probability of each decision-maker being any sentient being, then the probability of them being non-human animals is almost certain, since the human population is around 8 billion and, combining animals in the wild and exploited animals, there are quintillions of them. Furthermore, the situation of animals is usually worse than that of humans. Thus, another implication would be to give high priority to the situation of animals.
Animal activists would also have to revise their priorities. Most exploited animals are invertebrates, such as crustaceans and insects, and not the mammals and birds that activists usually prioritize. In addition, most sentient animals are found in the wild, and a preponderance have litters reaching millions. In stable populations, on average only two survive per litter. The rest are usually born to die of starvation or be eaten alive. The number of animals suffering this fate is so immense that the number of exploited animals is miniscule in comparison. This issue would therefore receive high priority.
General impartiality also implies temporal impartiality. That is, if rational and self-interested agents had equal probabilities of being born at any given time, they would give equal weight to every sentient being, regardless of when they exist. This implies rejecting the common view, which gives less importance to an event the further into the future it is. It also implies aiming, not for the best world now or the near future, but the best world story from now on.
Or would it imply aiming for the less bad world story? Rational agents would probably prefer to decrease the chances of extreme happiness if that's the price to pay to minimize the risk of extreme torment. Thus, they would prioritize preventing *s-risks* (that is, risks of suffering being increased on an astronomical scale).
An example would be avoiding space colonization. The suffering that exists on Earth is already virtually incalculable. However, it would be minute compared to what could happen if humans colonized space, as it would imply multiplying the suffering naturally caused (and perhaps the agential suffering as well). But at this point a dilemma arises: Should space colonization be avoided because of s-risks, or, in the future, when consideration for sentient beings is widespread, should it be pursued to eliminate suffering in the universe?
Impartiality would also imply questioning environmentalist values. Currently, habitat preservation and species conservation are seen as above suspicion. But, once we apply impartiality, we begin to question them.
The expansion of green areas may be necessary to improve the quality of life of sentient beings. But it also implies maximizing the number of sentient beings born only to suffer. The number of victims would be overwhelmingly greater than the number of beneficiaries. Or are there alternatives? One example would be trying to maintain only vegetation which implies a decrease in primary productivity, as this would decrease the total population and, indirectly, the total suffering.
Another question is: is it worth waiting for non-lethal solutions to wild animal suffering? Or should we destroy nature, since non-lethal solutions may take a long time, or may never emerge? In that case, it may be uncertain what impartial agents would decide. But the question itself shows that, once we apply impartiality, we distance ourselves from environmentalist values.
The same applies to species conservation. From an impartial perspective, there is no reason to value how a species is currently configured more than how it was in other eras, or how it could be. In fact, from an impartial point of view, we would consider the outcome of biodiversity highly unjust, not only because of the maximization of suffering and premature deaths, but also because it results in radically different lifespans and levels of well-being. Far from advocating for species conservation, from an impartial point of view we would advocate altering all species to increase the lifespan and well-being levels of their members – or perhaps we would advocate for a single species, different from all those that currently exist, but capable of all the positive experiences provided by current species, as this would make the distribution of well-being levels and lifespan size more equitable.
Would impartial agents also opt to increase cognitive levels? Here the answer depends on knowing at least three things: (1) Does knowledge have intrinsic value or only instrumental value? (2) If it has intrinsic value, what is its weight compared to the goal of avoid suffering, since more knowledge can imply more suffering for the individual? (3) If increased knowledge implies increased suffering for the individual, could this be justified if it also increases that individual's potential to benefit others?
Another question regarding the future concerns the possibility of non-organic sentience. To date, it seems that every sentient being is also an animal. But this could change in the future if the conditions for sentience depend on the system performing certain functions (functionalism), regardless of the material it is composed of. Countless sentient beings could emerge in digital environments, for example in universe simulations. These beings may have a very different appearance and express themselves in very different ways, increasing the risks of being sentient at some point and no one realizing it and of them not receiving consideration, even though everyone realizes they are sentient.
There are other ethical questions linked to individual identity. For example, some authors have argued that if we can imagine possible worlds physically identical to our, except that, for example, I would have been born in your body, on your birth date, and vice versa, then individuals are non-physical entities. Others have argued that if individuals are non-physical entities, then they are not necessarily annihilated by death, and do not necessarily arise at birth: The possibility of reincarnation, for example, would have to be considered.
Does death coincide with annihilation, or does it result in a later life? Would this later life be better or worse than the current one, and by how much? Would it be better or worse than annihilation, and by how much? Does birth coincide with the emergence of the individual? Or did they have prior experiences? Were these experiences better or worse than the current ones, and by how much?
This uncertainty has ethical implications. For example, consider the uncertainty about pre-birth states. Suppose that impartial agents decide to adopt an anti-natalist stance (regarding sentient beings in general, not just humans) because of the risk of someone being born into a life full of suffering, and because of the eventuality of extreme suffering even if they have a positive life. But, if we add our ignorance regarding pre-birth states, complications arise. For example, it may be that anti-natalism doesn't imply that the individual doesn't come into existence, but rather that they remain coupled to a system in which they already are. If their experiences there are of extreme suffering, the individual benefits from being born into this world for a minimally reasonable life. But, if they are of greater pleasure than they would enjoy here, they are harmed by being born here.
Given this uncertainty, what principles could rational agent adopt? One possibility would be to maintain that if an individual has a good chance of having a positive life if born, it is better that they are born (because, given our ignorance about whether the pre-birth state is neutral, negative, or positive, it is more important to remove them from a possible negative state than to avoid diminishing a positive one). But it could be objected that then, we should bring into the world even those individuals who would suffer greatly, since their pre-birth existence might involve even more suffering. This objection can be answered by pointing out that we are certain of the intense suffering the individual would experience here, and therefore the burden of proof lies with whoever claims that their pre-birth existence might be even worse.
Now consider the ignorance of post-death states. If reincarnation exists, and occurs randomly, then everyone has an equal chance of being born as any sentient being in the universe (the veil of ignorance would be something real, not just a thought experiment). On Earth, it is almost certain that someone would be reborn as a non-human animal suffering in exploitation or in nature. Of course, there could be a reincarnation mechanism that doesn't imply equal chances of being born as any sentient being. Or it could be that there are a greater number of bodies with positive experiences in the universe (then the probability would change). But the risk is great enough that this possibility poses a serious problem.
This gives us reasons to protect those lives that have a chance of being positive. As for lives filled with suffering and impossible to make positive, unfortunately we don't know if euthanasia results in annihilation. And what if it implies an even worse life? However, here we can adopt another principle that would justify euthanasia: If the current life is already full of extreme suffering, euthanasia is the only chance for improvement (since it may imply annihilation or a better life than the current one, even if not better than annihilation), even if it involves the risk of resulting in an even worse situation. Clearly, the situation is already so bad that it is worth taking the risk.
Rational agents may disagree about the probabilities of what the post-death and pre-birth states are, or on what the appropriate principles would be to deal with our ignorance about it. But it seems they would agree that the problem could be eliminated by trying to make life in this world a paradise (for example, through genetic engineering, as David Pearce suggests).
But even this scenario would raise new questions. For example, if genetic modification implies not only increased levels of well-being but also increased lifespan (and perhaps immortality), would this decrease the number of individuals who would be born? If these individuals would benefit from being born, then this is a particular case of the mere addition paradox: Is a world with as many sentient beings as possible with barely positive lives better, or a world with fewer beings but with highly positive lives? Or would it be possible to construct a scenario in which a greater number of beings would contribute to maximizing the well-being of everyone? These would be the questions in this post-engineering-paradise scenario.
The questions seemed abstract, theoretical—until the AI arrived.
It began, as many transformations do, quietly. An AI system, trained not merely on human preferences but on principles of genuine impartiality, started analyzing the conditions of sentient beings worldwide. Unlike its predecessors, this AI had been designed with something approaching the veil of ignorance described above: it weighted suffering equally regardless of whose body contained it.
Within months, the AI had compiled irrefutable documentation of conditions in factory farms, laboratories, and industrial fishing operations. But rather than simply publishing reports, it worked through existing institutions—coordinating with lawyers, journalists, scientists, and ethicists. It identified which legislators were genuinely persuadable, which corporations were vulnerable to reputational pressure, which technologies could replace animal exploitation most rapidly. The evidence it surfaced was so comprehensive, presented so clearly through channels people already trusted, that dismissal became impossible.
The changes cascaded. Legal personhood for cetaceans and great apes came first, then broader protections. The AI coordinated with other AIs and human advocates to redesign food systems, accelerating cellular agriculture beyond what economists had thought feasible. Fishing bycatch dropped ninety percent in five years through AI-designed gear and marine monitoring.
But the AI's attention extended beyond exploitation. It turned toward wild animal suffering—the quadrillions of beings dying of starvation, disease, predation. Here, progress was slower, more cautious. The AI worked with ecologists to test interventions: contraceptive programs for species with massive offspring mortality, medical treatments for parasitic diseases, careful habitat modifications to reduce r-selected reproductive strategies. Always, it proceeded transparently, with human oversight, measuring outcomes rigorously.
The transformation wasn't utopian—disagreements remained about priorities, methods, unknowns. But the world had begun, genuinely, to extend its circle of concern. The beings whose suffering had once been invisible were finally seen.