Aug 15, 2024 08:00:00

Philosophy is important for the development of AI, and philosophers can gain new ideas from AI.

The AI era is expected to accelerate further, with a Google software engineer and a

Zapier co-founder running a contest to award up to $500,000 (approximately 78 million yen) to developers of intelligent general-purpose artificial intelligence (AGI) , and OpenAI, which created ChatGPT, predicting that a 'superintelligence' that exceeds human capabilities will appear in 10 years and allocating 20% of its development resources. In such an AI era, not only science and technology but also 'philosophy' is contributing to the development of AI, and a philosophy professor explains that this will continue to be important in the future.

Philosophy is crucial in the age of AI
https://theconversation.com/philosophy-is-crucial-in-the-age-of-ai-235907

According to Anthony Grayling, a professor of philosophy at Northeastern University London, philosophy has made important contributions to AI since its birth. For example, the 1956 Logic Theorist , known as the 'world's first AI program,' is a program for proving mathematical theorems using propositions such as those shown in the 1910 book ' Principia Mathematica ' by philosophers Alfred North Whitehead and Bertrand Russell. 'Principia Mathematica' is a highly regarded book that derives mathematical truth from axioms and inference rules in symbolic logic, and is considered 'one of the most important and original works in philosophy, mathematics, and logic since Aristotle.' Logic Theorist proved 38 of the 52 theorems in the book.

In addition, Grayling points out that the logic on which early AI focused relied heavily on fundamental discussions by mathematicians and philosophers. For example,

Gottlob Frege, a German philosopher from the late 19th century, is known as the 'father of mathematical logic and analytical philosophy' and developed modern logic that is also important for AI logic. Kurt Gödel , a logician and philosopher from the 1930s, is known for his research on the completeness theorem and incompleteness theorem, which are important fundamental theorems in the foundations of mathematics and computer science.

As mentioned above, the history of AI development is closely related to philosophy, but Grayling says that philosophy also plays an important role in modern AI based on deep learning. For example, ChatGPT, which generates natural conversational sentences, is trained on a dataset of billions to trillions of data, tracking and utilizing statistical patterns for each language. This is close to the ideas of linguistic and analytical philosophy of 20th century philosopher Ludwig Wittgenstein , and Grayling points out that ' Philosophical Investigations ,' which is called the most important philosophical book of the 20th century, contains important insights on how to analyze AI.

In addition, some philosophers believe that questions such as 'Can AI acquire consciousness?', 'Can AI 'understand' the language it processes?', and 'Can AI that generates images and text be creative?' are beyond the scope of science and require the help of philosophy. Margaret Bowden, a British cognitive scientist who also studies the philosophy of AI, argues that 'AI can generate new ideas, but it is difficult to evaluate the work in the same way as ideas created by creative humans,' and predicts that only an architecture that uses both deep learning and logical philosophy will realize AGI.

When OpenAI announced its prediction that 'AGI will begin to emerge in the next 10 years,' it was asked about the role of philosophy in the age of AI, and suggested that 'philosophy can help ensure that the development and use of AI is consistent with human values.' Grayling argues that if the adjustment of AI is a serious problem, it should be solved not only as a technical problem by engineers and technology companies, but also as a societal problem, incorporating the opinions of philosophers, social scientists, lawyers, policymakers, and ordinary citizens.

Finally, Grayling also touches on the possibility that AI may influence philosophy. Gottfried Leibniz, a 17th century German philosopher, suggested that we might see the emergence of ' computational reasoning machines ' that would derive answers to philosophical and scientific questions like an oracle. Not only has philosophy contributed to the development of AI, such as AI encoding logic to enable value-oriented evaluations that are completely free of human subjectivity, and there is a project to simulate the influence of social media information to see how we should form opinions, but it is also thought that the advancement of AI may significantly change the way philosophers think and give answers that have not been available before.