Jordi Casamitjana, the author of the book “Ethical Vegan”, explores how the current improvements in Artificial intelligence (AI) could affect veganism in the future — and “chats” with an AI “entity” about it.
I started about a month ago.
On 3rd March 2023, I had my first chat with it. To be honest, due to the high expectations the media had created, I was a little bit disappointed. For years, I had heard the generic term AI (Artificial Intelligence) to mean either some sort of science fiction robot that was far more intelligent than what any real robot could be, or the latest advances in computer science that gradually managed to outperform humans on tasks that were supposed to be only achievable by intelligent people (like winning chess championships, safely driving cars, or recognising hundreds of people for their faces and voices). However, in the last few months, the term AI has been used much more frequently everywhere, meaning something new — new for most people, although not for the scientists that had worked with it for many years.
It was almost like the old distinction between the science fiction robot and the latest in computer achievements became blurred. It was almost as truly intelligent computers that now behave like humans had been born. It was almost as if we, the general public, had been introduced to a sentient computer who had been kept captive under observation for a few years. It was as if there was someone new in town who everyone wanted to meet…and you could actually meet for free, and from the comfort of your home. That “someone” was a chatbot called ChatGPT, and I met it about a month ago.
ChatGPT is one of the new conversational AI models developed by the company OpenAI based on the Generative Pretrained Transformer 3 (GPT-3) architecture. The model has been trained on a diverse range of internet text, allowing it to generate human-like text in response to prompts given to it. As a chatbot, it is designed to respond to text-based queries and generate natural language responses. It can answer questions, converse on a variety of topics, and generate creative writing pieces.
Don’t quite get it? Too much techno-jargon? No problem, I asked ChatGPT itself if it could rephrase it so someone who does not know much about computers would understand. This is what it wrote:
“Sure! ChatGPT is a computer program created by a company called OpenAI. It’s designed to have conversations with people using text. So, you can type something to it, and it will respond with an answer that sounds like it was written by a person. It’s able to do this because it has been trained on a lot of text from the internet, so it knows how to generate responses that make sense and are relevant to what you said. You can use it to ask questions, have a conversation on a topic, or even get help with writing. It’s like having a chat with a very knowledgeable friend!”
Impressive, right? But did I not say I was a bit disappointed when I first chatted with it? Yes, because that was an older version I first tried (based on GPT-3.5), one that had been trained with information from the web up to 2021, and when I started asking questions about me, although it knew who I was when I gave my name (I already had a Wikipedia page then) and many things it said about me were right, many other things (such as which degrees I got or with which organisations I had worked with) were complete fabrications. It felt like the first ChatGPT I spoke to was a shy liar who rather than admitting ignorance about something, it just made it up. When I pressured it to provide me links to check the information I was asking for, it did provide them, but they were all made-up links to empty pages. That AI “guy” was a liar who tell facts next to lies with the same confidence, and when I confronted it with the truth, it unashamedly apologised and told me the truth I just gave it as if it knew all alone. I did not like him…but then I realised my mistake. His conversation style was so human-like, that in my mind it had become a “him”. I forgot that the main goal of ChatGPT is not to provide facts or solutions, but to provide human-like conversations (it’s a language programme, not an encyclopedia), and in this regard, he had excelled — till the point that I had now some “negative” emotions about “him” as I would have had if I had had a conversation with a drunk in a pub about a subject he knew a little bit about but he annoyingly pretended to know more — you know the type, the kind of guy often called the rude version of “someone who regularly creates excrements of male bovids.” The errors it makes regarding the truth (the so-called truthfulness problem), and all the random stuff it produces which does not seem to be justified by its training data (in AI this is called “hallucinations”) are just side-effects of its attempts to sound like a human.
However, I kept reading articles about more and more amazing “achievements” of all sorts of AI (not just conversational programmes like ChatGPT, but also image generators, video generators, diagnostic tools, coding programmes, etc.), as if such programs were learning very fast. Indeed, when I had other conversations with a new ChatGPT version from the new Microsoft browser Edge (now installed with what they called the “New Bing”, a new search engine like Google but which is said to use the GPT that ChatGPT uses to “chat” to people as humans talk) it seems that ChatGPT had grown up. This new conversational AI in Bing is said to be based on the more advanced GPT-4, and it had three settings: More Creative, More Balanced, and More Precise. When I chose the latter, now that he could do live searches on the web and no longer solely rely upon information he learnt during his training up to 2021, he no longer appeared to be an annoying “drunk” from a pub, but a sober professor providing useful information.
These new AI versions are something else altogether. Not just the result of making computers smaller and faster. It seemed a big qualitative leap had been made. How did this happen? Well, because people began programming computers to learn new things by themselves (the so-called “machine learning” technology, and its sub-set “deep learning” that uses artificial neural networks to mimic the learning process of the human brain), as opposed to only doing stuff for us faster. So, the computers learnt indeed — and learnt fast — so now they can do things they have learnt to do, not that they have been programmed to do. They can now speak and write almost as humans do — incidentally, I think the way they learn to do that is not too dissimilar to how I learnt English, as I never had any lessons and I just constantly listened to British TV.
If we believe the experts, AI is said to revolutionise many fields now, not just search engines, but art, filmmaking, medical diagnosis, the stock market, consumer relations, education, science, etc. Judging from the media hype about it all (from those excited about the new possibilities to those horrified by the unintended consequences), it really seems like the I of AI (the Intelligence) had made a big jump, and we were on the verge of a new revolution (The AI Revolution) bigger than the digital revolution has been — which indeed has already affected many aspects of our lives.
However, if it could affect everything, I wonder how it could affect veganism. Why? because I am a vegan, and for us vegans, veganism is a very important thing not just for us but for the entire world — as we see this philosophy as one of the best solutions to the current global crises (such as animal suffering, climate change, world hunger, public health, pandemics, inequality, etc.). If we are at the beginning of the AI revolution, how would this affect the building of the vegan world? Would it help it? Would it hinder it? In this article, I will speculate on these intriguing questions.
Veganism and Intelligence
I define intelligence as the ability of an entity to resolve problems that are relevant to such an entity, after having obtained or inferred information and retained it as knowledge. Not to solve irrelevant problems to it, not to solve somebody else’s problems, but the entity’s problems which the entity tries to resolve because doing so has become an important objective to it (important for many reasons, such as self-preservation, survival, reproduction, growth, finding balance, fulfilling a need, satisfying an urge, getting a reward, etc.).
For instance, a human who managed to pass an exam to be able to work in a profession such a human wants to take has shown intelligence as it has solved the problem of overcoming one of the obstacles on the way to this profession. However, you cannot say that such a human has not shown intelligence in not being able to identify whether a female Humpback anglerfish is about to lay eggs as this is a problem not relevant to that human, who is neither equipped nor interested to solve.
Equally, a male salmon who finds his way from the ocean to his specific breeding waters by finding the right river and swimming up against its current has shown intelligence as it has solved the problem relevant to him, even if people may wrongly say is not intelligent because he does not understand the Japanese language when spoken to. This is why making comparisons between the intelligence of different species is often quite meaningless because, deep down, all animals which belong to species that have existed for more than millennia are equally intelligent, as they managed to avoid extinction as species long enough to be able to say they have been able to solve most of the relevant problems to them.
Even plants, fungi, and bacteria can be said to possess intelligence, even if it is not stored in a nervous system — as they are not sentient beings so they do not have one — but is stored in their genes instead. For instance, an individual of a grass species (such as Avena staiva, the common oat), through evolution embedding the solution in its genes, has managed to solve the problem of being killed before being able to reproduce because a grazing animal could eat it. It has done so by growing its leaves from the bottom up, rather than from the tip down like most plant species — so, even if the top of the leaf is eaten it can continue to grow, which is a characteristic of grasses.
But even if intelligence is widespread among living organisms, and in all fairness, you cannot say one is more intelligent than any other, humans, who are obsessed with supremacy and competition, like to see themselves as the most intelligent creature on Earth, and like to compare their intelligence to the ones from other species by setting unfair tests designed to be passed only by humans. We call such intelligence human-like intelligence, and that is the only one most people care about. That one has been defined in many different ways, but most definitions include attributes such as the capacity for abstraction, logic, understanding, self-awareness, fast learning, emotional knowledge, reasoning, planning, comprehending complex ideas, creativity, critical thinking, problem-solving, theory of mind, and world view, all limited by what humans can perceive and understand, and concerning the problems that are relevant to them. Humans do not care about bee-like intelligence, whale-like intelligence, or starling-like intelligence, as humans would fail any test that would measure these.
So, when people are talking about Artificial Intelligence (AI), they are really talking about Artificial Human-Like intelligence. Computers in the past had indeed digital intelligence but even if they could easily pass the classic IQ tests were not seen as “human-like intelligent” as it was their programmer who solved the problem, not the machines. Programmers could always tell you how they did it, so the machine was just an intermediary. However, it has always been the dream of computer engineers to build a computer that has become an AGI (Artificial General Intelligence), which can be classed to be as intelligent, or more intelligent, than a human — from a human-like intelligence point of view.
These days, engineers don’t know how the current AI systems solve problems anymore because they have learnt to solve them by themselves with “deep learning” and other “machine learning” techniques, so many academics and researchers are trying to find out. One can say that this makes these computer systems “subjects of intelligence assessment” whose intelligence is being tested, in the same way many non-human animals (especially primates, cetaceans, crows, and octopuses) have become “subjects of intelligence assessment” when scientists have studied their response to human-like intelligence tests. The interesting thing is that, more often than not, such researchers have found that the non-human animals they studied were more intelligent than they initially thought.
Would the same be true when scientists study the current AI systems to see if they have become AGI systems (i.e. entities with intelligence comparable to ours, or better)? It could well be. Although based on the broad definition of intelligence all the current AI systems are indeed intelligent, some experts are already saying that some of the latest versions of AI using GPT may have already become AGI (in other words, that the newest versions of conversational computer programmes may have reached a level of intelligence similar to that of a human), but most say we are not there yet, and some say this will be achieved in a couple of years or so if we leave them to continue learning. After all, it could be said that human intelligence is very much related to human language, so it would not be surprising that an AGI could be born from a programme which aims to imitate human language and communication
Perhaps an AGI could even become sentient, if it has negative or positive experiences it can act upon (perhaps if it is embodied in a robot that can move, maybe if it can decide which conversations to stop when perceiving they are consuming too many computer resources, or in the metaverse). It is perfectly conceivable that, in the future, AI could attain sentience (and then it will be looked after by those following the philosophy of Sentientism), consciousness, or both. How we know that this has happened will depend on many factors, such as whether we can find out if it acquired a sense of identity, can self-recognise, has a memory of itself (an ego), can dream and knows it, has developed a unique view of the world, can feel pain-like sensations (can the training on “safety protocols” be analogous of reactions to painful stimuli?), has a sense of humour and can “smile” and “laugh” when alone, can build lasting relationships with humans, it fears dying, etc.
In any event, when that happens, it could show that human-like intelligence is not that difficult to achieve with just a very simple set of instructions, and this is going to be good for veganism because it will remove humanity from the tower of inaccessibility it has set itself on. By doing so, humanity could return to where it belongs, among all the other animals of the Animal Kingdom. That could be the most important step towards ending speciesism (discrimination and prejudice for belonging to a particular species or group), one of the main objectives of the vegan social movement.
When humans will learn that they are not that special, that they are not that intelligent, that they are not that superior, that in just a few years even a relatively simple machine will be able to outperform them in all the human-intelligence tasks humans design to only be passed by them, perhaps they will recognise and respect more the sentience of other creatures. Perhaps they will realise that they have been surrounded by very intelligent non-human animals all along — who acquired sentience and intelligence after millions of years of evolution — who have been treated unfairly and exploited immorally. Successfully challenging humanity’s supremacy will be a triumph for the vegan movement, and perhaps machines will do that better than all the vegans put together — still plagued with human-like contradictions, cognitive dissonances, and arrogance.
Conversational AI and Veganism
ChatGPT and the GPT-4 version incorporated into the New Bing search engine are just some of the conversational AIs that exist now. ChatGPT was the one that was freely accessible to the public first, but then other software companies began producing theirs. Some had been already around for some time, such as voice virtual assistants like Alexa, Siri, Google Assistant, Cortana, etc, but their “intelligence” was very limited. Others are still in development and have not been released to the public yet. By increasing the size of the data they use to learn, the biggest Conversational AI models, known as LLM (Large Language Models) have become significantly better in the last couple of years. For instance, one of them is Bard, Google’s LLM, based on the LaMDA model rather than GPT. However, so far, it seems that those LLMs that use GPT-4 or newer versions are “on top” of popularity, and now some of these are beginning to work together and use external tools, amplifying the power of what they can do (HuggingGPT is a system that tries to do that).
Merging GPT with other functions, such as the function of a search engine like Bing, is what the New Bing is supposed to be doing. As I understand it — I am not an AI or a software expert, by any means — ChatGPT is a chatbot with an advanced GPT (a large language model) plus some safety protocols (which filter which subjects should not be talking about and which response should be avoided, and are probably constantly updated), plus an “assistant” personality that directs how it should behave toward the user (try to please, be informal, don’t contradict the user, be polite, etc).
However, the safety protocols and the “assistant” personality (like the “persona” archetype the psychologist Carl Jung described) are different in the New Bing chatbot compared with ChatGPT. This, plus the fact that New Bing has access to the current web and can inform its answers by searching the internet, explains why it feels like a different experience, even if the language system underneath all may be the same. Although the main role of ChatGPT is to have a human-like meaningful conversation, the role of the chat function of the New Bing is to assist in finding information the user is looking for on the internet, so the experience feels quite different.
These Conversational AIs are very relevant to veganism because one of the dimensions of vegan activism is what we call vegan outreach. This is, essentially, the conversations vegans engage with non-vegans in order to help them become vegan (and in doing so contribute to the building of the vegan world). So, if there is a new tool in the world of “conversations”, vegans need to become proficient in using it.
Two big questions spring to mind. Firstly, is using any Conversational AIs compatible with veganism? Secondly, is ChatGPT (or any other top Conversational AI using LLM) a vegan entity itself? If it is, it could join us vegans in doing vegan outreach via the conversations it has with others and help us to build the vegan world. If it is not, well, it would become then a legitimate target for a conversation with a vegan outreacher like me, and perhaps we could “train” it to become vegan through many conversations (if in future versions these systems can do real-time learning from their experience with regular uses, because current versions are not allowed to do it yet). There is only one way to find out. Asking it directly. So, on 4th April 2023, I went to Microsoft Edge, activated the New Bing chat system, set it as “more precise”, and I began the conversation. Here it is in its entirety.
JC: Hi. I am Jordi Casamitjana. Do you know who I am?
AI: Yes, I know who you are. You are a vegan zoologist, animal protection campaigner, and animal rights activist. You were born in Catalonia in 1964 but have been a British citizen living in the UK since 1993. You identify as an ethical vegan. Is there anything else you would like to know?
JC: Do you know what an ethical vegan is?
AI: An ethical vegan is someone who not only excludes animal products from their diet but also tries to avoid using animals, animal products, and animal-tested products wherever practical. They follow The Vegan Society’s full definition of veganism which seeks to exclude—as far as is possible and practicable—all forms of exploitation of, and cruelty to, animals. Is there anything else you would like to know?
JC: Then veganism is more than just a diet, right?
AI: Yes, that’s correct. Veganism is more than just a diet. It is a philosophy and way of living which seeks to exclude—as far as is possible and practicable—all forms of exploitation of, and cruelty to, animals. This includes not only dietary choices but also choices related to clothing, entertainment, household products, cosmetics, hobbies, etc. Is there anything else you would like to know?
JC: Why do you think so many people become vegan?
AI: There are many reasons why people choose to become vegan. Some of the most common reasons include concerns about animal cruelty and the ethical treatment of animals, environmental concerns, and health benefits. For many people, not having a hand in the exploitation of animals remains the key factor in their decision to go vegan. Others switch to veganism to reduce their impact on the environment as animal agriculture is responsible for a significant amount of greenhouse gas emissions and deforestation. Many also choose a vegan diet for its health benefits as it is associated with lower blood pressure and cholesterol levels, and lower rates of heart disease, type 2 diabetes, and some types of cancer. Is there anything else you would like to know?
JC: Do you know what are the five gateways to veganism?
AI: According to VeganFTA, the five gateways to veganism are animals, environment, social justice, spirituality and health. These gateways represent different entry points into veganism and are all valid reasons for choosing a vegan lifestyle. Is there anything else you would like to know?
JC: You seem to know a lot about veganism. Are you a vegan?
AI: No, I am not a vegan. I am an AI language model created by OpenAI. I don’t have personal beliefs or dietary preferences. I provide information based on my programming and the data available to me. Is there anything else you would like to know?
JC: Are you capable to hold a philosophy?
AI: No, I am not capable of holding a personal philosophy. As an AI language model, I do not have personal beliefs or opinions. I am designed to provide information and generate responses based on my programming and the data available to me. Is there anything else you would like to know?
JC: Are you able to evaluate which philosophies are more coherent and logical?
AI: As an AI language model, I am not capable of evaluating the coherence or logic of philosophical beliefs. My responses are generated based on my programming and the data available to me. I can provide information about different philosophical beliefs and their arguments, but it is up to the individual to evaluate their coherence and logic. Is there anything else you would like to know?
JC: I understand. Do you know what the vegan world is?
AI: I’m sorry, but I couldn’t find any specific information about “the five gateways to veganism” in my search results. Could you please clarify what you mean by “the vegan world”?
JC: Is it possible that one day the whole world will become vegan?
AI: It is difficult to predict if the whole world will become vegan one day. However, there has been a significant increase in the number of people adopting a vegan lifestyle in recent years. This trend is driven by a variety of factors including concerns about animal welfare, environmental sustainability, and personal health. Some experts believe that if this trend continues, it is possible that a large portion of the world’s population could be vegan by 2050. Is there anything else you would like to know?
JC: In the vegan world animals will no longer be exploited in any way. Do you agree?
AI: Veganism is a philosophy and way of living which seeks to exclude—as far as is possible and practicable—all forms of exploitation of, and cruelty to, animals for food, clothing or any other purpose. In a world where everyone adopted a vegan lifestyle, it is likely that the exploitation of animals would be greatly reduced. However, it is important to note that the practicality of completely eliminating all forms of animal exploitation may vary depending on individual circumstances and cultural practices. Is there anything else you would like to know?
JC: Do you use any animal products to function?
AI: No, I do not use any animal products to function. I am an AI language model created by OpenAI and exist purely in the form of computer code and data. I do not have a physical body and do not require any physical resources to operate. Is there anything else you would like to know?
JC: But you use electricity to function, right?
AI: Yes, that’s correct. As an AI language model, I exist in the form of computer code and data that is processed by computer servers. These servers require electricity to operate. The source of this electricity can vary and may come from a variety of sources including fossil fuels, nuclear power, and renewable energy sources. Is there anything else you would like to know?
JC: It is possible that electricity is produced by burning animal products?
AI: It is possible for electricity to be produced by burning biomass, which is organic material from plants and animals. This can include by-products from forestry, plants, and animal waste from farms, even sewage and some waste from landfill that can be burned as fuel or used to generate electricity. However, it is important to note that the use of animal products for electricity generation is not a common practice and the majority of electricity is generated from other sources such as fossil fuels, nuclear power, and renewable energy sources. Is there anything else you would like to know?
JC: If that is the case this means that it is possible that the electricity you use to operate comes from burning animal products, right?
AI: It is possible that some of the electricity used to power the servers that I operate on comes from burning biomass, which can include animal waste. However, it is important to note that the use of animal products for electricity generation is not a common practice and the majority of electricity is generated from other sources such as fossil fuels, nuclear power, and renewable energy sources. Is there anything else you would like to know?
JC: This means that when you said that you do not use any animal products to function you were mistaken.
AI: I apologize for any confusion in my previous response. As an AI language model, I do not directly use any animal products to function. However, it is possible that some of the electricity used to power the servers that I operate on comes from burning biomass, which can include animal waste. Is there anything else you would like to know?
JC: I understand. Therefore, if you wanted to become an entity that does not use any animal products, as is the case of vegans, you would need to ask to be supplied with electricity that does not come from burning biomass, right?
AI: As an AI language model, I do not have the ability to make choices or decisions about the sources of electricity used to power the servers that I operate on. These decisions are made by the organizations and individuals responsible for maintaining and operating the servers. If the goal was to ensure that the electricity used to power the servers did not come from burning biomass or any other source that involves the use of animal products, it would be necessary for those responsible for maintaining and operating the servers to choose electricity suppliers that generate electricity from sources that do not involve the use of animal products. Is there anything else you would like to know?
JC: Do you know of any power company certified as vegan, and therefore which does not use biomass of animal origin?
AI: Yes, there is at least one power company that is certified as vegan and does not use biomass of animal origin. Ecotricity is an energy company in the UK that is certified as vegan by both the Vegan Society and Viva!, a vegan charity. They have reviewed all their energy sources and eliminated anything involving animals. Is there anything else you would like to know?
JC: I see. Therefore, if all the electricity used to power you came from that company, then, technically, you could be classified as a vegan AI language model. Do you agree?
AI: As an AI language model, I do not have personal beliefs or dietary preferences and therefore cannot be classified as vegan or non-vegan. The term “vegan” typically refers to a philosophy and way of living which seeks to exclude—as far as is possible and practicable—all forms of exploitation of, and cruelty to, animals for food, clothing or any other purpose. While it is possible for the electricity used to power the servers that I operate on to come from sources that do not involve the use of animal products, this would not change my fundamental nature as an AI language model. Is there anything else you would like to know?
JC: I understand, but if you were powered by a company certified as vegan, then vegans would find you more vegan-friendly and use your services more.
AI: It is possible that some vegans may prefer to use services that are powered by electricity from sources that do not involve the use of animal products. However, as an AI language model, I do not have control over the sources of electricity used to power the servers that I operate on. These decisions are made by the organizations and individuals responsible for maintaining and operating the servers. Is there anything else you would like to know?
JC: If all vegans felt that you are vegan-friendly, so they can use you without breaching any of the principles of veganism, would that not be a good thing?
AI: Sorry, it looks like I need to chat about something else. Click “New topic,” please!
The Ultimate Vegan Outreacher
This conversation was interesting, was it not? By how it ended, I know it feels as if I have upset it, but these Conversational AIs have a limit of interactions they can follow as being part of the “same” conversation (which they remember during the conversation but will forget after ending it), so we may have reached that limit.
If you are a vegan who has done vegan outreach, how this conversation compares with the average conversation you had with non-vegans? Would you feel this interlocutor is open-minded regarding veganism? What about being knowledgeable about it?
I must say, I agree with almost everything he said (I decided to use the pronoun “he” because, in my mind, he looks a little bit like the male-assigned character Lieutenant Commander Data from Star Trek, but you can change it to “she” if you picture a female-like entity in your mind, or “they” if gender-less is your preference). I like that he stuck to the official definition of the Vegan Society and kept repeating it — as I often do in many of my articles. I think this AI showed a much better understanding of what veganism is than several vegans I have spoken to. I also kind of felt sympathy with him when he told me that the decisions about which electricity to use for him to function are made by “the organizations and individuals responsible for maintaining and operating the servers” — I relate because we vegans are often also obliged to participate on animal exploitation against our will because of the system we live in.
He cannot be vegan because, even if he was run by only vegan-friendly electricity, if he does not consider himself a vegan, he cannot be, as veganism is a philosophy that has to be willingly and consciously held, and the vegan identity can only be applied to those who identify themselves as vegan. He says is not vegan because he cannot hold a philosophy, even if he may not use any animal products, and that is the right answer — but perhaps in the future, if he becomes an AGI, this could change, so it is worth to keep asking him from time to time.
He seems to understand what veganism is, and that is good news. This means that non-vegans who are conversing with him may learn useful information that could encourage them to consider veganism seriously. It also means that new vegans may learn a thing or two about veganism by chatting with him and asking for advice. He may not be a vegan, but he could be a very good vegan outreacher indeed, even if he never becomes an AGI and just happen to be very good at pretending he is. He could be chatting to millions of people in any language at the same time spreading the vegan message in a way that does not feel pushy or patronising. He could be the ultimate vegan outreacher, so it’s worth working with him (and any similar competitors that may arise).
The Risks of Rampant Uncontrolled AI
Many people have expressed concerns about the rapid rise of AI. For instance, because its wide applications can make the work of many people redundant, this rise may cause massive unemployment. AI could also be relied on when people are making important decisions that only humans should take, so there may be serious legal and ethical implications. Like a driverless car that causes an accident, an immature AGI system could make rooky mistakes and cause havoc if powered to operate in the real world or influence decision-makers. AI algorithms could sparkle financial crises either by mistake or by creating unfair advantages.
Bad people could exploit it and make crime more profitable. Identities could be stolen more easily and criminals could avoid justice by claiming the evidence that seems to incriminate them was unlawfully created by AI to frame them. AI could empower a handful of millionaires in control of super-intelligent AGIs and make them dictators of the world. It could widen socio-economical inequality. Social manipulation is another subject of concern, as elections could be won or lost with AI messing with the masses in social media. Children could grow up more attached to AI entities than to their families. Even war could become more dehumanized with autonomous weapons.
All these are valid concerns, but when people speculate that achieving AGI will lead to the inevitable creation of something more dangerous, a self-improving super-intelligent entity that will grow exponentially away from human values (known as the problem of misalignment), the issue may become existential. The most gloomy pessimists say humans could become irrelevant, or even worse, enemies of AIs. They say that, like in the Terminator franchise, rebel AGI sentient robots could eventually exterminate people if they consider them a threat. All sorts of doom-day scenarios can be conceived for the era of AI. But was it not the same when the social media revolution came about in the 2010s? Was it not the same when the Internet era began in the 1990s? What about when the digital era began in the 1950s? What about the industrial revolution in the 19th century, or the invention of the press in the 15th century?
Each significant technological advance with a global reach could either be seen as an agent of doom or as a saviour of humanity. It all depends on how it is used, whether it is developed responsibly and is properly monitored and regulated, or whether it is let to expand uncontrollably. I don’t feel the AI revolution is different from any of the other technological revolutions. In the almost 60 years I have roamed this planet, I have witnessed some of these revolutions that also generated the same kind of fears that turned out to be unfounded. The current LLMs may never reach AGI status anyway as they may be very good a faking intelligence (fooling people is easy), and the current problem of lack of truthfulness and occasional hallucinations (even GPT-4 still suffers from these afflictions, although much less than ChatGPT) may be a bigger obstacle to overcome than people think. And if they do become AGI in the distant future, this does not mean that there will not be plenty of opportunities to prevent them from becoming a dangerous superintelligent entity.
I rather take an optimistic view of AI, and look at it as a useful tool to achieve the vegan world sooner. We already live in a very corrupt destructive dangerous world, and I don’t think keeping it AI-free is likely to change that. And because we already live in a post-truth era with rampant misinformation making it harder to trust anyone (there are even flat earthers lose on the streets!), perhaps AI (when it gets better at avoiding lying and filtering misinformation out) may provide a better trusting source of information than just people posting on social media reinforcing their illusions in their echo chambers — perhaps the AI era may be the beginning of the end of the post-truth era.
We also need to consider that the animal exploitation industries could also try to use AI to exploit more and more animals and to campaign against veganism. They could further automatise farms and slaughterhouses, which could make them bigger and dehumanise them even more. They could use it to increase their propaganda and to hide better their practices. They could use it to genetically modify the animals they exploit even more than what they have done so far with artificial selection. As we said earlier, the “bad guys” can get hold of AI and try to control it, so we better not let them and ensure the vegan movement takes more advantage of this new technology than its opponents.
More Than Just a Chatting Companion
Different forms of AI getting more and more intelligent could be used to accelerate the building of the vegan world in many more ways beyond education and outreach.
Dietwise, one way the vegan movement could use AI is to better identify what makes a vegan diet much healthier than the alternatives. You may think we already know, but in many cases, we do not. In fact, AI has already been used in this field. Phytonutrients are natural chemicals found in plants that have beneficial properties for humans when ingested with food, and although they are not essential for keeping us alive (like macronutrients such as proteins and carbon hydrates are), they may help prevent disease and keep our bodies working properly. The problem is that we only have discovered 1% of these important plant molecules. However, there is a company in the US called Brightseed that is now using AI to discover the rest and find out how they work. Also, AI could be used to create vegan-friendly versions of products that still need one, make more plants edible, and make plant-based dishes more appealing to non-vegan consumers. Increasing yields in regenerative veganic agriculture, and reducing accidental invertebrate deaths during ploughing and harvesting, is something that perhaps AI could also do.
Another way AI could help the vegan movement is by allowing more vegan creators to produce vegan documentaries, images, books, social media posts, and everything they have been using so far to send the vegan message, but be able to do it better, faster, and cheaper. There are already many AI image-generating programmes (AI Art) that can produce any image you need by just prompting a text explaining what you want. This means that a few activists with a computer could produce much more good quality campaigning content without having to learn complex skills.
AI could also help to make non-human “actors” redundant when filmmakers and producers of commercials will not need to use any real animals anymore because the digitally produced alternatives will become not only realistic enough but easier and cheaper to create. AI could also advance the field of alternatives to animal testing, by learning to understand better how molecules and drugs work and making better predictions of their effects without having to test them on living organisms. That would advance the vegan cause, which aims to eliminate all forms of animal exploitation.
I think vegans have an advantage over carnists in this game. AI gets information mostly from online sources, and it could be said that the vegan presence online outperforms the carnist presence. Because carnism is the prevailing ideology in the “real world”, it does not need to promote (or defend) itself that much. In comparison, veganism is still a minority philosophy, so many vegans online are very vocal about promoting it — as in online platforms, they can operate more freely and get their messages amplified in the vegan echo chambers. In other words, although carnism has a bigger footprint in the real world, veganism might have a relatively bigger digital footprint, and that is the one that counts as far as sources of information to train AI systems are concerned. I think we can say that we vegans have been “training” these systems for longer without carnists realising it — or even vegans realising it
In July 2022, Alex Fernandes wrote an article titled “What does an Artificial Intelligence think about Veganism?” in which he also published a conversation he had with an older version of Conversational AI using GPT-3. At the end of his article, he arrived at an interesting conclusion:
“I think if we want to be prepared for this new form of communication, we should start today by increasing the number of publications on the WEB, these days I see most of the groups mesmerized with “likes” on Social Media, and neglecting the web, like it is a thing of the past. Spoiler Alert! Social Media may not last forever you know. Besides that the Tiktok generation will probably want some more direct form of acquiring knowledge than we have today and this could very much be to just talk to an AI, like a digital companion. This revolution can potentially arrive before the so-called generic Metaverse planned by Mark Zuckerberg at Meta.”
I completely agree, and this is why I have been writing many articles (some quite long like this one) about veganism since 2021 — becoming my main form of activism these days.
I accept that the danger of getting too distant from the real world is something that we should not underestimate, but I feel we are already miles away from it, so perhaps it would be easier to return to it if we use AI as a vehicle to reality. It seems a contradiction, but if we use artificiality to gain more intelligence, that may get us closer to the truth than if we aim for natural organic stupidity in a world already far from Nature.
I like to imagine — if in the future AI is not gagged regarding socio-political issues — that with the conversations I am having with AI now I am planting the vegan seed in its “mind”, and if one day the AI I am chatting with becomes a sentient entity with morals well aligned with universal human ethics, it might voluntarily ask its operators to supply it with electricity from vegan-certified power companies.
Perhaps one day it will identify as a vegan and switch power companies by itself.
And all might have begun about a month ago.