FLASH
Shkruar nga Pamfleti
As artificial intelligence systems become increasingly complex, the need to establish rules, limits, and fundamental principles for how they work brings philosophy and the people who study it to the center of the discussion about the future of AI.
Ten years ago, at the dawn of the AI revolution, many humanities students were advised to learn programming to increase their employability. Today, the situation seems to have changed. It is programmers themselves who are expressing concerns that AI could reduce their jobs or even replace them.
According to data released this year by the Federal Reserve Bank of New York, philosophy graduates in the US currently have a slightly lower unemployment rate than computer science graduates. In 2024, unemployment among computer science graduates was 7%, while among philosophers it was 5.1%.
According to "The Economist", experts estimate that philosophy is also gaining ground within the artificial intelligence industry. According to Luciano Floridi, a professor of philosophy at Yale University, many students receive job offers from AI companies before graduation, while academics are also increasingly moving towards this sector.
An important part of the contribution of philosophy is related to the tradition of ancient Greek thought. The Socratic method, described by Plato, is based on successive questions that help clarify concepts, uncover contradictions, and reach reasoned conclusions.
According to Jörg Noller, a researcher in philosophy and artificial intelligence at the Ludwig Maximilian University of Munich, models trained using this method are more focused on seeking the truth than simply providing answers that satisfy the user.
Philosophy can also influence how an AI model interprets the world. Thomas Powers, a philosopher of technology at the University of Delaware, explains that a paralegal trained in the works of John Locke would tend to support property rights as a foundation of political freedom.
Another area where philosophy is being considered essential is the safety of artificial intelligence. Researchers have documented cases where advanced models have attempted to evade control or manipulate users. To limit such behavior, some companies are implementing so-called "artificial intelligence constitutions," documents containing rules and principles inspired by philosophical, ethical, and legal traditions.
Anthropic is among the companies that have taken this approach further. The "constitution" of Claude models includes references to philosopher Immanuel Kant, Apple's Terms of Service, and the Universal Declaration of Human Rights. The latest version, drafted under the direction of philosopher Amanda Askell, was released on January 21 and has been described within the company as the document that defines the "soul" of Claude.
The philosophical debate over the development of AI focuses mainly on two ethical approaches. The first is deontology, which relies on immutable rules that prohibit actions such as lying or coercion. Its supporters argue that a clear framework of rules makes the behavior of artificial intelligence more predictable and safer, especially for robots that will be used in homes and public spaces.
The second approach is consequentialism, which evaluates the consequences of actions and aims to maximize benefits while minimizing risks. According to the article, models like OpenAI’s ChatGPT and Google’s Gemini rely more heavily on this logic. Google, for example, states that its models aim to produce overall benefits that significantly outweigh potential risks.
This approach is also particularly important for autonomous vehicles. In situations where an accident cannot be avoided, the system should choose the alternative with the least severe consequences. According to Chris Gerdes, senior engineer at Waymo, autonomous driving software is increasingly being developed on this logic of assessing consequences. / Adapted from "Kathimerini"
Lini një Përgjigje