[…] When asked about concerns regarding the powers of machine learning, Dr. Kissinger expresses concern that in the realm of AI, “We have no great philosophers” which he calls “an unprecedented challenge for humanity.” We are reminded that the most successful scientific revolutions and intellectual movements were guided by philosophical insight. Thus, philosophers will be best-positioned to meet these new challenges.

[…] Ethical implications of AI, the strain on resources from increasing GAI workloads, and intellectual property rights are a few among many of the pressing issues surrounding AI today, often perceived as too vast to tackle effectively. Even more concerning, Vogt and Haas observed that computer scientists, engineers, and software developers tend to be less inclined than philosophers and artists to approach these questions. Enter the ValuesLab.

Vogt and Haas are co-founders of the ValuesLab—a project that originated from their reflections on gaps in AI discussions and research that philosophy could bridge. In late 2023, Haas launched the ValuesLab website to highlight the most important considerations for advancing the project. As the two analyzed ongoing AI projects and trends, they realized that “one thing in particular is missing: a place that puts philosophers in a position where they actively contribute to AI.” Haas states, “This is the ValuesLab’s mission. We want to help shape new projects, by collaborating with researchers who build AIs, inside and outside of academia. We also want to keep things focused and independent.”

[…] Most important to its mission is the ValuesLab’s use of the Socratic model to explore critical questions relevant to computational intelligence. Vogt explains that being “Socratic” involves being “driven by questions and a love of inquiry. It also means that some of the most important values relate to truth and understanding” which she explains are critical to tackling challenges related to comprehending how Explainable AI systems reach certain conclusions.

Moreover, Socrates is famous for asking “What is X?” questions, where X is a certain unknown. Vogt and Haas agree that this question type is critical for understanding AI. Vogt asks, “What, for example, is AI? Some experts say it’s a probability calculus. Others see a future where we interact with machines that have minds.” Haas continues: “The Socratic method distinguishes between straightforward factual questions, where the answer is a piece of information, and questions that require inquiry. The ValuesLab is interested in the latter. In that context, the best answer can be a question, one that launches a new line of inquiry. For example, one new player who pursues this kind of idea in exciting ways is perplexity.ai.”

[…] If philosophy is crucial to AI, some might ask whether computer scientists might simply enroll in an applied ethics course. Vogt argues that this approach would be inadequate, stating that “AI needs philosophy, not merely applied ethics.” That is, developing AI necessitates drawing insights from philosophical inquiry in various subjects including language, value, equity, bias, and autonomy—topics that extend far beyond what a single course can cover.

This team-taught, interdisciplinary class covers the history of AI, the development from Neural Networks (NNs) to Large Language Models (LLMs), philosophy of AI, as well as the role of AI in music and writing. Four sessions are devoted to foundational philosophical questions that bear on AI. Session 1: Can we ascribe beliefs and intentions to AI? Can LLMs speak? Can they lie? Session 2: Does human cognition work like an LLM, with next-token prediction? Session 3: Can AI be aligned with human values? Session 4: How does probabilistic next-token prediction compare to memory-based AI? How does AI affect human memory?

The philosophy of AI is an emerging field. Right now, AI is importantly concerned with LLMs. It is also concerned with the relation between natural and artificial intelligence. Researchers and public discourse ask whether AI can be “aligned” with values. Accordingly, key questions in Philosophy of AI relate to language, thought, and values. 

The seminar starts with the widely debated alignment problem (Part I: weeks 1-4). Independent of how alignment works, it is by no means clear what the desired outcome is. People disagree about values. With which values should AI be aligned? At times the answer is: with “human values.” Are “human values,” in this context, the different and incompatible sets of values human beings have? If yes, what about the values that we should have?

AI researchers often focus on fairness, typically understood as the elimination of bias (Part II: weeks 5-6). We examine relevant notions of fairness and ask how fairness relates to other values, including and especially accuracy.

Next, we ask whether it makes sense to ascribe beliefs and intentions to AIs (Part III: weeks 7-11). Can AIs engage in reasoning, lie and be held responsible? We discuss “explainable AI,” asking whether AI-outputs can be understood. 

Finally, we examine questions about language as they apply to AIs (Part IV: weeks 12-14). How can LLMs cope with famously tricky components of language and thought, such as generics and implicature?

Part of the seminar are workshop sessions associated with the ValuesLab. Invited guest speakers come from a range of fields. This seminar aims to contribute to dialogue between philosophers and AI developers, and to a shared vocabulary.

Katja Maria Vogt, a Professor of Philosophy and PI on the Values Lab, who moderated the session, opened by asking not what fairness meant but how fairness came to take the spotlight in cultural conversations about ethical AI. She posited that “fairness” is often used in place of “justice” is as a way to ground discussion and sidestep big philosophical questions, a hypothesis drawn from a John [Rawls] paper that posits that fairness is considered more of a workable notion than justice. 

Vogt suggested that maybe there is no way around the big picture questions and interdisciplinary contexts like the Data Science Institute play a crucial role in helping society pose them.

“No matter what our students major in, they’ll need the analytical tools to ask what AI is and how we should relate to it,” says Katja Vogt, professor of philosophy at Columbia and co-founder of the ValuesLab. “These questions require collaboration between fields, including philosophy.”

“As a philosopher, I’m interested in values, language, and the mind. Applied to AI, this means I’m interested in what it would mean for AI to be aligned with values and whether this is possible,” she explains. “In class, we discuss this with regard to a range of values — some ethical or moral, some concerned with language and thought: fairness, truth, accuracy, understanding, interpretability, and more.”