photo_camera Jo-Anne McArthur / We Animals
Cows In A Sale Yard
Longtermism and Animal Farming Trajectories
Michael Dello-Iacovo
Strategy Lead and Researcher
December 23, 2022

Edited by Jacy Reese Anthis. Thank you to Janet Pauketat, Ren Springlea, and Yip Fai Tse for providing useful feedback.


The end of animal farming will be the end of farmed animal suffering. However, humanity may exist for many eons to come, and the way animal farming ends may have substantial impact on that future. In this post, I ask three questions from this longtermist[1] perspective:

  1. What sort of trajectory towards the end of animal farming are we currently on (i.e. social or technological)?
  2. Can we change the trajectory of how animal farming ends?
  3. What would different trajectories of the end of animal farming imply for future issues?[2]

The strategic implications of this post are:

These arguments have slightly updated me towards thinking socially driven trajectories are better for long-term outcomes than technologically driven trajectories, and towards thinking that the trajectory is an important strategic consideration.

Which trajectory are we on?

Some argue we are set to end animal farming, perhaps by 2035 or 2100.[3] The median estimate by users of Metaculus, a popular prediction platform, for when the number of concentrated animal feeding operations (CAFO) in the US will decline to 90% of their 2020 levels is 2067. Metaculus users also predict a mean probability that slaughterhouses will be banned in the UK by 2050 of 10%.

There has been substantial but limited progress towards this goal to date. Polls conducted by the Vegetarian Resource Group over the past three decades found that 3–4% of adult Americans report vegetarian eating habits (e.g. never eating “meat,” “poultry,” or “seafood”) and 1–2% report vegan eating habits, which increased up to those rates by 2000 but has remained roughly steady since then, though with variation in survey wording that makes establishing a trend difficult. Gallup found in 2018 that 5% of American adults self-report as vegetarian and 3% as vegan, which has also remained steady since their first data collection in 1999. The estimated proportion of vegans and vegetarians even within a single country varies greatly between surveys due to differences in sampling and wording, such as social desirability bias and different ways of asking for self-reported diet, which is also more prone to inaccuracy than direct measurement (e.g., purchase history).[4]

The proportion of animal rights activists is more difficult to find an estimate of over time but also appears to have increased substantially, such as the budgets of animal rights organizations. The growth and development of plant-based food technology and cellular agriculture has been substantial in the past decade. The market share of plant-based milk alternatives grew by 9% from 2017–18, and as of 2018, plant-based milk represents 15% of the total market share of milk. Plant-based product sales generally grew twice as fast as overall food sales in the US in 2021. A marketing firm, MarketsandMarkets, estimates that the plant-based meat market globally is expected to grow 99% from 2022–27, though such estimates are extremely speculative and often lack transparent methodology.

There are indications that social change has led to substantial discomfort with and opposition to the status quo: 49% of Americans agree with the statement, “I support a ban on the factory farming of animals”; 47% support a ban on slaughterhouses; and 33% support a ban on all animal farming. Rethink Priorities found 20% support and 16% support, in two similar studies, for making slaughterhouses “illegal” even with a critical framing of, “Supporters of this policy say that slaughterhouses should be banned because it is wrong to kill animals…”  which further declines to 8% in a treatment condition when participants were told they will have to explain their choice. In general, support for institutional change seems much higher than making analogous behavior change at the individual level. 

With these figures in mind, the end of animal farming could be driven primarily by social change, such as a rise in the moral consideration of animals, or by technological change, such as the introduction of realistic animal product alternatives that are cheaper and more environmentally friendly than animal products.[5][6] There are many ways in which social and technological change can, have, and will work together, but advocates inevitably face trade-offs between them in many marginal scenarios, such as deciding which organizations to fund or which jobs to take. There are many actions we can take that intuitively fit as both social and technological, and as with most dichotomies, it can be conceptualized as a multidimensional gradient, though a detailed embedding is beyond the scope of this post. Figure 1 shows some examples.

Figure 1: Venn diagram showing the overlap between social change and technological change. The specific categorization is debatable; this is to give a general idea of the taxonomy, rather than argue for the placement of particular actions.

Overall, it is difficult to say what trajectory we are currently on. We likely won’t know until we are closer to the end of animal farming and can say with some confidence what factors contributed most to that success. Both social and technological factors are certainly at play. If the social and technological trends continue at their current rates, my overall impression based on experience in the field is that we are currently on a technologically driven trajectory towards the end of animal farming.[7] This is supported by slow, if any, growth of vegetarianism, and the rapid growth of plant-based food technology such as Beyond Meat and Impossible Foods and cellular agriculture such as Upside Foods. However, this does not account for interactions between social and technological factors, such as increased access to animal product alternatives making social change easier.

Social change factors

Most stories of moral progress are described as socially driven trajectories, such as the civil rights and women's suffrage movements. Animal farming may go through a similar shift, but there are numerous disanalogies between such movements, such as that animals can’t advocate for themselves in the way many oppressed humans have.

Several social changes already seem to be improving attitudes towards animals, such as an increase in urbanization and pet ownership. The ratio of cats and dogs to humans in the US increased from 22% to 50% from 1967 to 1988. In The End of Animal Farming (2018), Jacy Reese Anthis writes that, “As wealth increased and horses made way for cars, city dwellers may have owned pets to show off their wealth or to fill the emotional gaps left by a lack of working animals, relatives, and close friends in the city.”

Other social change factors may include spillover from areas unrelated to animals, such as increased concern for other humans. Steven Pinker in Better Angels of our Nature (2012) argues that violence is declining over time; this trend might not hold with violence towards animals in agriculture, though it may hold in attitudes towards animals not used as commodities such as companion animals.

Technological change factors

Numerous moral victories appear to have been won primarily through technological change, such as the automobile and other technological advances largely replacing the use of horses in transport and farming (Levin 2017) and the replacement of wood heaters with modern heating reducing the number of deaths due to air pollution (Xue 2017). The consumption of animal products may go through similarly rapid decline.

Plant-based food technology and cellular agriculture may end animal farming by displacing animal products. This seems plausible due to the inefficiency of animal products, such that more efficient ways of producing the same or similar end products may win out for economic reasons.[8] It is also possible that such efficiency gains will not persist, leading to recidivism back into animal agriculture.

Simply eating animal products reduces attribution of mind to animals (Bratanova et al. 2011; Bastian et al. 2011). This suggests that replacing animal farming through technology may have an effect on people’s attitudes towards animals, reducing cognitive dissonance and making it easier to include animals in one's moral circle. In other words, social change may be achieved through technological change, kicking off a virtuous cycle of progress.

However, a Sentience Institute study found that awareness of animal-free food technologies (AFFT) seems to have a negative impact on animal farming opposition. While this is a single study and thus very limited evidence — further research is necessary — possible explanations for this include:

A particularly important technological factor is artificial intelligence (AI), given increasing AI capacities and the likelihood of human-level machine intelligence and an AI takeoff by 2100 that may be transformative in human history. AI technologies can be applied to make animal farming more efficient, such as predictive modeling and optimization of animal product output, or to make animal-free food, such as predictive modeling and optimization of combinations of plant ingredients to make animal product alternatives or the media for cultured meat bioreactors. Andrea Owe and Seth Baum (2022) write about the potential for weak AI today and strong AI in the future to harm animals in their broader discussion of the importance of nonhumans in AI ethics, a topic also addressed by Singer and Tse (2022).

Can we change the trajectory?

We can partly answer this question by looking at how past moral victories were achieved (e.g. primarily through social change or technological change) and estimating the effect that trajectory had in the long run. Assessing the counterfactual impact of events in history is extremely difficult, but it can provide some useful insights. That topic is discussed extensively in Jamie Harris’ (2019) “How Tractable is Changing the Course of History?”, and an illustrative example is Luke Muehlhauser’s estimate of the industrial revolution’s impact on human wellbeing. This type of analysis examines “trajectory change”, which is the “bend in the curve” of moral progress, or any long-lasting or permanent change to the value of the world.[9]

Harris addressed this question in much more detail in his blog post, “How Tractable is Changing the Course of History?” (2019). He compares the role of three sorts of forces in historical events: indirect or long-term factors, contingencies of luck and hard-to-influence decisions, and the actions of thoughtful actors. Scholarly opinion on the causation of historical events supports the idea that thoughtful actors can have some control over the course of history.

An example argued for by Will MacAskill is that the end of slavery might have taken place decades or centuries later — or not have occurred at all —  if the Quakers hadn’t played such a pivotal role in the early stages of the abolitionist movement (MacAskill 2022).

Another example explored in Sentience Institute’s report on the British anti-slavery movement is the argument that the industrial revolution played the primary role in ending slavery through making slave labor no longer commercially viable. Eric Williams defended this claim in 1944 in his doctoral thesis Capitalism and Slavery, though this view seems less popular today. One notable rebuttal is that the number of slaves and the economic output of a slave were still rising when it was abolished. Seymour Drescher claims that the UK suffered a 5% loss of GDP due to the end of British slavery and the subsequent drop in sugar production (Drescher 2010), which makes an economic cause for the end of slavery in the UK less unlikely. Further, MacAskill claims that mechanization of agriculture in the US south didn’t take place until the 20th century, which would make it implausible as the proximate driver of US abolition.

Even when thoughtful actors steer historical trajectory in the short run, there may be a rubber banding of moral value back to its original direction or all the way back to its original value.[10] The below graphs show this. Consider a scenario in which thoughtful actors have been able to create some moral progress (i.e. some trajectory change has taken place; some moral value has been created). In Figure 2, the value eventually reverts down to the original trend. This could be considered as position rubber banding. In Figure 3, the function value does not revert to the original trend, but the slope does, i.e. slope rubber banding[11]. In both scenarios, there is value (i.e. an area between the different trajectories). However, in Figure 2, the benefit is temporary, while in Figure 3, moral value will be counterfactually higher even after the trajectory returns to its original slope, leading to an impact that scales as the future extends further. One can also consider higher-order rubber banding, such as reversion to the original second derivative (i.e. slope of the slope) even if the first derivative (i.e. slope) never returns to its counterfactual.[12]

Figure 2: Position rubber banding. The blue zone shows the counterfactual difference caused by the moral progress that starts at time point 3 and reverts by time point 9.

Figure 3: Slope rubber banding. The blue zone shows the counterfactual difference caused by the moral progress that starts at time point 3 and reverts in slope by time point 6.

For example, position rubber banding could occur if a campaign to switch egg production to cage-free hen housing lead to an equal and opposite industry backlash that reverts the cage-free switch within 6 years. In this case, while some hens may have lived better lives during those 6 years, no hens are living in better housing due to this campaign after 6 years. In this case, other moral progress (e.g. cage-free campaigns in other countries) would continue in either scenario, leading to a positive slope in either scenario. Slope rubber banding could occur if the cage-free campaign inspires campaigns in other countries, and that inspiration is met with equal and opposite reversion, but the original cage-free switch is not reverted. In that case, the progress will continue to increase moral value accrued over time by those hens living better lives, but the value produced in year 10 would be no more than the value produced in year 7. Of course, these are oversimplified examples meant only to illustrate the model.

Moral progress for farmed animals could rubber band, including recidivism from an animal-free food system back to animal farming (i.e. position rubber banding)[13] or accelerated moral progress not spilling over to other areas such as artificial sentience (i.e. slope rubber banding). Different trajectories may have different tendencies to rubber band.

At first glance, it doesn’t seem like there has been position rubber banding of moral progress because well-known progressions have persisted even to this date, such as most enfranchisement of increasingly diverse human populations.[14] Arguably the largest change in humanity’s trajectory, the industrial revolution, has been followed by largely persistent economic growth at least until the 1960s (e.g., the Great Divergence), suggesting that rubber banding is uncommon.[15]

Insofar as rubber banding is a major concern, the codification of moral progress in law and other sticky institutions is a higher priority. For example, there has been recent success in corporate campaigns to make companies commit to source their eggs from cage-free providers. If there were a high probability that this progress would rubber band back to the previous state,[16] advocates should prioritize solidifying this progress with changes in legislation, which will likely be easier to achieve now than after the progress reverts. Such campaigns can leverage the interest of early adopters of improved animal welfare practice to ensure that non-adopters are forced to comply through legislation (Carey et al. 2020).

While codifying progress in law would make the progress stickier, it would not preclude recidivism. A recent example is the overturning of Roe v Wade[17] by the US Supreme Court in 2022 that reverted what many saw as landmark moral progress made in 1973. In general, individual legislative changes can be part of a broader rising tide, and rolling back the entire tide of legislative changes would be more difficult than rolling back individual legislative changes.

Codifying moral progress in law is a sort of value lock-in,[18] events where the state of the world becomes hard or impossible to change. Many events can be viewed as lock-in, such as human extinction, creating a new technology that would otherwise not exist, and spreading technological or cultural practices to other planets when interplanetary contact is limited, especially when the growth of a bubble of colonization reaches physical limits.

Value lock-in is particularly likely through von Neumann probes. If it were to ever occur, it would be difficult to change the practices in other star systems, and once interstellar colonization begins, it would likely continue at an exponential rate in an expanding bubble of space. From this point, the ability of a single actor or even group of actors to change the social or technological practices of interstellar colonies could be effectively zero. This could occur if animal farming were brought to other planets[19] or if interstellar planets are terraformed[20] — thus spreading wild animal suffering.

Creating a new technology is lock-in insofar as it tends to be less reversible than social change, in the sense that technologies are rarely extirpated from the body of human knowledge (Anthis and Paez 2021). Depending on the technology, this could be positive lock-in, such as animal-free food, or negative lock-in, such as transformative AI that is aligned with harmful values. Technologies can become obsolete or are banned for concerns such as safety, as in the cases of leaded petrol and hydrofluorocarbons,[21] as well as true eradication of knowledge during catastrophes such as the deliberate and accidental destruction of libraries.

The counterfactual of technological progress often seems to be a mere delay in its introduction. For example, would the light bulb have been invented anyway if Thomas Edison and other inventors hadn’t developed it in the 1870s? Given the number of other inventors who had been independently working on similar technology, such as Joseph Swan developing a light bulb using carbonized paper filaments in 1860, it seems likely that a similarly good light bulb would have been developed eventually, and perhaps even within a few years.

How will the trajectory affect the future?

Predicting the impacts of technological  and social change on moral outcomes is difficult due to the unpredictability of complex systems. While it seems intuitive that the advent of fossil fuels reduced the need for whale oil and therefore reduced whaling, sociologist Richard York (2017) argues against this. He claims that, while fossil fuels were a replacement, their larger impact was the development of modern whaling technologies that allowed for cheaper hunting, killing, and transport, as well as new whale products being developed, such as hydrogenation allowing for whale oil-based margarine. Instead, he suggests whaling only declined when this explosive growth in whaling decimated the population.

Similar arguments on more recent technologies include the development of electronic storage mediums having significantly less impact on paper consumption in offices than expected (Sellen and Harper 2003) and the generation of non-fossil fuel energy typically having less effect on displacing fossil fuel energy than expected, also argued by York (2012).

Similarly, while replacing horses in transport and farming is generally seen as a victory, it had potential downsides. Because horse riding still exists (e.g. as entertainment), this victory being won through technology instead of social change might have been a missed opportunity, just as ending factory farming with technology may be. It is of course difficult to say what the counterfactual situation would have been, and it’s possible that in the absence of this technological change, horses would still be used for transport today, just as factory farming may persist into the future despite its inefficiency.

A norm of waiting for technology for moral victories could make people less inclined to take moral action in the future without technology making it easier, analogous to going vegan before Beyond Burgers and similar plant-based meats were available, because they expect it to be easier later. This may be particularly true for cellular agriculture as a norm of waiting for technologies better in every way (e.g. taste, cost, health), which may set a particularly strict norm for addressing future moral catastrophes. While a similar argument can be made in favor of technological norm-setting, future work building technologies are probably less promising than future moral actions.


The end of animal farming will probably have spillover effects on other areas of moral concern, such as recidivism of animal farming, artificial intelligence and sentience, wild animal suffering, and conflict and cooperation, and the way that animal farming ends will probably affect that spillover.

Generally speaking, I expect social change to have better spillover effects than technological change because improved moral attitudes towards farmed animals are expected to expand the moral circle more broadly. Technological change causing moral spillover is more dependent on specific circumstances than social change, such as the technology having specific impact on other areas of moral concern. Spillover of technological change also seems more likely to be negative than spillover of social change.

Spillover to recidivism of animal farming

On the individual level, it seems to me that ethical reasons for veganism will lead to longer adoption of a vegan diet than health and environmental reasons. Arguably this is evidenced by survey data suggesting 58% of former vegetarians/vegans say they were motivated by health, 27% by animal protection, and 22% by concern for the environment, compared to 69% of current vegetarians/vegans saying they are motivated by health, 68% by animal protection, and 59% by concern for the environment.

On a wider, societal level, animal products being seen as healthier or more environmentally friendly may lead to widespread messaging not broadly focused on the nonhumans themselves, leading to a less persistent post-animal farming future because health and environmental circumstances can more easily change. However, as many like Tobias Leenaert have argued, people may adopt ethical reasons for veganism later at a greater rate after becoming vegan for health or environmental reasons (Leenaert 2017). Just as this may happen at an individual level, it may happen at a societal level, which could mitigate the upside of animal-centric messaging.

Spillover to wild animal suffering

Effective altruists have been increasingly focused on helping wild animals, who vastly outnumber even farmed animals and suffer from both human and natural sources. The same considerations for spillover concern to future farmed animals and artificial sentience (better precedent, faster progress) apply to wild animals, but the naturogenic suffering makes the issue more complicated.

For example, public acceptance of a consequentialist or deontological message that we should leave animals alone and preserve the environment may lead people to think that exploiting farmed animals is wrong, but it also might lead them to think that we shouldn’t intervene in the lives of wild animals as we would intervene to help humans in need. In contrast, public acceptance of a primarily utilitarian message may have more emphasis on suffering and preference frustration, which may lead to more acceptance of the idea that wild animals suffering naturally is still bad and a problem we should address. However, adoption of a naive version of utilitarian ethics may have downsides such as excessive interventionism or myopic optimization of complex natural environments. This is not to say that the naive versions of other ethical theories don’t also have downsides, such as a libertarian view that attempts to preserve wild animal autonomy by avoiding intervention even though that could curtail more autonomy by leading to lives constrained by the harsh natural environment.

Impact on artificial intelligence and spillover to artificial sentience

Jamie Harris has written on the importance of artificial sentience, including arguments that digital minds are likely to vastly outnumber biological minds in the future due to many advantages such as self-modification and longevity (Sotala 2012).

One of the main considerations relating AI to the end of animal farming is value alignment.[22] In the case where animal farming ends before a value-aligned AI takeoff, the trajectory of the end of animal farming may affect the values with which the AI is aligned. A primarily social change trajectory that has led to widespread increased concern for animals may carry over to AI having more concern for animals — and sentience more generally — compared to a primarily technologically driven trajectory.

In the case where an aligned AI takeoff occurs before animal farming ends instead of after, we might expect a greater chance that animals will continue to be exploited in animal farming, albeit more efficiently, because the alignment will happen with less animal-concerned input values. This argument suggests ending animal farming as quickly as possible, regardless of those other effects such as better historical precedent from socially driven change.[23]

Impact on conflict and cooperation

Consider a scenario where the end of animal farming is not uniform, e.g. some countries end animal farming while other countries continue business largely as normal. This could lead to conflict or reduced cooperation on other multilateral stakeholder issues (such as nuclear armament) and exacerbate other risks. Currently, no country has progressed along this trajectory enough for this to be an issue. It’s unclear how strong the effect of countries differing on animal farming would be compared to other differences that already exist, such as political/economic systems (e.g. democracy, capitalist, socialist), climate policy, etc. These other differences may dominate.

There are countries with historically low consumption of animal products, such as China, but this has generally not been for ethical reasons. As these countries develop economically, their animal product consumption generally increases quite rapidly. India has arguably had the most influence of cultural reasons for eschewing animal products with 23%–39% of the country being vegetarian, primarily as part of Hinduism. It is not obvious that this has been the cause of any increased conflict with their neighbors. There is tension and conflict between India and Pakistan, but differing views towards animals does not seem to be the cause. Conflict between Hindus and Muslims around the Hindu sacredness of cows seems to be the strongest example of farmed animal conflict, but ultimately the animal aspect seems more of a manifestation of conflict than a source of it.[24]

Animal welfare has also been a critique leveled against other countries. Examples include western countries criticizing the dog meat industry in some Asian countries,[25] Australians criticizing live export of animals due to the animal slaughter practices in other countries (e.g. halal slaughter; Pendergrast 2015), and other countries criticizing Australia for their slaughter of kangaroos. It seems that none of these cases have led to significant action, with the most notable exception being perhaps the 1986 moratorium on whaling by the International Whaling Commission on its members.[26]

Ending animal farming through technology may have less risk of sparking conflict insofar as social change would be more entangled with extant social conflict. However, animal-free food technology taking longer to spread through low-income countries may lead to conflict across wealth brackets (Reese 2018). A possible solution to this problem is for high-income countries to play a role in helping low-income countries advance without doing the same damage that wealthier people have, as has been proposed for climate change solutions and could form a part of those efforts (Romani and Stern 2011).

Perhaps the most notable example of major domestic conflict driven by a moral divide is the US civil war. The Confederacy seceded from the US to protect slavery, rather than the Union starting the civil war to end slavery. The Union arguably went to war to hold the nation together, rather than specifically to end slavery. This is supported by Lincoln stating that they had no intention of ending slavery in his inaugural address in 1861. Arguably this was somewhat a foreign conflict because the South requested support from France and Britain, but they refused to intervene. Thus, the US civil war evidences the ability of moral progress to effect major domestic conflict.

Future research questions

[1] Longtermism is the view that a key factor in moral action today is its impact on the long-term future. Strong longtermism states that the far future is overwhelmingly important. We should consider the effects of different trajectories on the risk of recidivism back into animal farming, the risks of other human-caused animal suffering, wild animal suffering, the treatment of artificial sentience, and conflict and cooperation between countries and other entities.

[2] Issues I consider are spillover to recidivism of animal farming, wild animal suffering, artificial intelligence and sentience, conflict and cooperation, and moral circle expansion.

[3] It's interesting to think about what the end of animal farming actually means (e.g. a 90% reduction in animals farmed, legal prohibition in every country), but it does not make a difference for the purposes of this blog post.

[4] For a summary of these polls, see Šimčikas (2018).

[5] For example, Singapore giving regulatory approval for Eat Just’s cell-based chicken. See also cell-based meat being cleared for human consumption by the US FDA.

[6] I define a historical event as “driven primarily by social change” if removing 50% of social actions taken to facilitate the event is more likely to prevent the event than removing 50% of technological actions, though this is still imprecise.

[7] As a counterpoint to this, there may be some limitation to cellular agriculture that prevents it from being made cheaply at scale, or most consumers may just not want to eat it. There is also harmful technological change, such as technology that allows for new or more efficient ways to farm animals, which is out of scope for this blog post.

[8] A counterpoint to this may be that there are some inefficient foods that may stick around because people just like them, such as coffee and chocolate. There is some supporting evidence in a study that found that consumers may still prefer animal products over cheap, highly meat-like animal-free foods, though this is based on reported preferences which may not reflect actual consumption. This may not be the most accurate comparison given that we expect realistic animal product alternatives to be more similar to their original than for coffee.

[9] See Nick Beckstead’s 2013 thesis for a more detailed discussion on changing the far future.

[10] The idea of rubber banding of moral value was initially suggested by Jacy Reese Anthis.

[11] These are simplified linear graphs to illustrate a point, but progress may be linear, sigmoidal, exponential, cubic, etc.. For example, some trends such as Moore’s Law have been exponential, and space exploration at a fixed speed expanding in all directions would be cubic. This also assumes that moral progress is upward. It might also be argued that moral progress has been flat or downward, which would have different implications for rubber banding.

[12] Thanks to Jacy Reese Anthis for raising these possibilities of different sorts of rubber banding.

[13] One recidivism scenario is using animal products as a food source during a global catastrophe, which has been proposed by ALLFED.

[14] This claim could be subject to a selection effect. When thinking of moral victories, it’s easy to only think of the victories that are still in place, which we are more likely to be aware of. Some possible counterexamples include the overturning of Roe v Wade by the US Supreme Court in 2022 affecting abortion laws, and the Islamic Revolution of 1979 affecting the rights of women in Iran.

[15] It has led to some negative outcomes, such as growth of industrial animal agriculture and harms of climate change, but these are not rubber banding per se in the sense that the harms are on different axes than the preceding gains; it is not as if the industrial revolution led to widespread reduction in animal suffering that was subsequently reversed by animal farming.

[16] For example, it seems likely that after a global catastrophe, humans would revert to animal farming standards such as using caged eggs in the interest of survival. If moral progress continues, this reversion should become less likely over time, though other factors like increasing technological power may outweigh this. One should consider whether reversion has in general become less likely over time.

[17] For more detail see this interview with Tracy Weitz.

[18] For more detail on values spreading see Tomasik (2013).

[19] Research is already being performed to examine the viability of this, including insect farming, fish farming, and animal agriculture.

[20] See Dello-Iacovo (2016) for more detail on terraforming and how it relates to lock-in.

[21] It’s worth noting that hydrofluorocarbons were invented to replace chlorofluorocarbons (which were invented by the same person who invented leaded petrol, Thomas Midgley Jr.), which were even more harmful.

[22] See Gabriel (2020) for more discussion on this.

[23] The same may be true for misaligned AI takeoff, though many misalignment scenarios involve the AI optimizing for some arbitrary value function, often caricatured as a “paperclip maximizer.” In that case, other trajectory differences may have little impact.

[24] Cows are a sacred animal for Hindus, who are the majority in India, but not for Muslims, who are the majority in neighboring Pakistan. However, conflict between India and Pakistan seems to be primarily driven by territorial dispute over Kashmir and cross-border terrorism. Within India, there are instances of violence as a result of differing views towards cows, such as a man being killed over rumors that their family was storing and consuming beef, a legislator being assaulted for hosting a beef party, and a clash between Hindus and Muslims over rumors of cow slaughter. Vigilante cow protection groups have also grown in number.

[25] See Chas Newkey-Burden’s article on The Guardian and Humane Society International for more detail.

[26] See the International Whaling Commission for more detail on the 1986 moratorium on whaling. While Japan was part of the IWC, in practice, Japan never stopped whaling and caught hundreds of whales each year through what they call research missions. In 2019 Japan withdrew from the IWC and resumed commercial whaling.

[27] To examine this, we could ask the question of whether we would be in a better position with respect to animal protection if we stopped using horses for moral reasons rather than for technological reasons.

Subscribe to our newsletter to receive updates on our research and activities. We average one to two emails per year.