Aristotle still is regarded as one of the most, if not the most important philosopher. His ideas have laid the groundwork for many modern philosophical thoughts, and I do not know any person seriously interested in philosophy who has not taken the time to dive into the works of Aristotle. To reveal how the thoughts of Aristotle can be linked to the current challenges we face with AI, I will dedicate the first couple of articles for this ‘Verum Machina’ project to the ancient writings and thoughts of Aristote.
I know that I do not know
One of the most wellknown sayings of Aristote comes from Plato’s apology, in which Aristote provides an apology1 before a crowd of people who will have to decide whether he is guilty of the charges against him. To be more precise, he is being charged by not recognizing the gods of the city and corrupting the thoughts of the youth. According to the oracle, Aristotle is the wisest of all. He himself does not agree and goes on a search for someone wiser than him. However, what is discovers during this trip is that there are many who call themselves wise and claim to have the answer to many questions. Aristotle, in contrast, knows that he does not know all. This thought alone should make him wiser than all who pretend to know things, or actually believe they know all.
Is AI all-knowing?
One thing is for certain and that is that AI has a large knowledge base. Depending on the model that you are working with, AI also has impressive reasoning capabilities. However, could one say that AI is wise? To be able to do so, one would have to know what ‘being wise’ would mean. Wisdom is often seen as a virtue, one of the well-discussed topics by Aristote. We will dive further into the notion of wisdom and virtues in one of the other articles about Aristotle but today we will remain at the question whether AI is all-knowing.
“The unexamined life is not worth living.” — Plato, Apology
Science in the eyes of AI
AI is often used as designer or conversational partner. It can be of great help to automate daily tasks or to perform tasks which we are rendered incapable of due to our human nature. However, the role that AI can perform in science is ambiguous. AI has shown to be of help in certain empirical sciences. It has been able to solve certain math problems which had yet to be solved or the added value of AI within many other field of research has already been proven. We should ask ourselves what the worth of science even is if not discovered by humanity, and sometimes not even understood. Due to industrialization, craftmanship has lost its beauty in many areas. Currently, we are at a dangerous point where a similar things might happen to the world of science.
The role of AI in questions that concern the metaphysical in science is even more interesting. Take for example phenomenology, the science of conscious perception for first-person view. As this brief explanation already indicates, one should eb conscious to be able to come up with questions and search the answers to these questions. Since one cannot research something that does not lie within grasp. When asking phenomenological questions to AI, it might explain that it is limited in this field due to the fact that it is not conscient. However, it might also go with the conversation as if it was. Taking the ground rule that AI is not conscious 2, we can draw the conclusion that AI does not know all yet could act like it does. This is very similar to what the what has been said in the apology.
Dreaming AI
When using AI, it somethings might feel like we are working with some type of all-knowing genie. Not just a genie that knows all, but also one that is capable of many things. As anyone who has been working with AI will definitely already have experience, AI is not prone to something called dreaming. Dreaming is when an AI model ‘creates’ new false information and uses it as a response to the user. It claims to know things and provides an answer to the question, yet whether he really does know remains ambiguous because the answer is false. When the AI would actually (be able to) know the right response, the situation becomes even more undesirable and risky.
Conclusion
The fact that AI cannot know all, think about the metaphysical, yet might act like it does exposes one of the main flaws of AI. AI cannot be wise and still we entrust it with many matters, some even question morality and ethics altogether. We as mankind should be careful that we are not idolizing AI and it’s big show. Because that is what is it. AI raising the impression that it is more than a tool is a potential danger that should be nipped in the butt. The link between sophist and AI can be made, and I dare say that Aristotle would agree with me on the points declared in this article.