Media

A public panel conversation hosted by journalist and broadcaster Samira Ahmed, with media artist Refik Anadol, composer Holly Herndon, and Dr Kathryn Eccles, exploring how generative AI and machine learning are reshaping authorship and creativity, and what remains uniquely human.

Panel discussion with Llion Jones (co-inventor of the Transformer architecture), exploring the ways in which current AI systems still fall short of human-like intelligence, and we can close that gap.

A podcast interview on the alignment problem for language models.

An interview for "Decoded" on CNN about AI and creativity.

A podcast interview about AI and creativity.

A long-form conversation on various topics related to my research, including the nature of self-consciousness, and the 'vector grounding problem' for laguage models.

"How do large language models work? What are the dangers of overclaiming and underclaiming the capabilities of large language models? What are some of the most important cognitive capacities to understand for large language models? Are large language models showing sparks of artificial general intelligence? Do language models really understand language?"

"Machine minds can work a paintbrush, but are they really making art? In episode 80 of Overthink, Ellie and David talk with guest Raphaël Millière, scholar and philosophy lecturer at Columbia University, on the aesthetic merits of computer-generated art. They discuss the thorny marriage of art and technology in everything from the early days of photography to YACHT’s AI-assisted pop songs. Why do we expect art to express human emotions? Is prompt-engineering for AI models an art in itself? And, if ‘great artists steal,’ is DALL·E the greatest artist of us all?"

"ChatGPT is great — but does it ‘understand’ what it is telling us? Raphaël Millière joins Vasant Dhar in episode 60 of Brave New World to understand on what’s going on inside ChatGPT — and the larger questions that arise from this."

"As AI becomes increasingly prevalent in our daily lives, it is important to understand the mechanisms behind its decision-making processes. What are the algorithms and models that underpin AI, and how do they differ from human thought processes? How do machines learn from data, and what are the limitations of this learning? These are just some of the questions we will be exploring in this episode. Raphaël will be sharing insights from his work in cognitive science, and discussing the latest developments in this rapidly evolving field. So join us as we dive into the mind of artificial intelligence and explore how it thinks."

A long-form conversation about various issues related to artificial intelligence and language models, including semantic competence, language understanding, compression, compositional generalization, systematicity, the Language of Thought hypothesis, conceptual combinations, the grounding problem, world models, and theory of mind.

A discussion with Ellie Pavlick (Brown University) about compositional generalization, referential grounding, and large language models.

An interview on CNN Chile, on the occasion of Congreso Futuro 2023.

"Chatbots are artificial intelligence algorithms designed to converse with humans. They have gotten so compelling in recent years that it is easy to misunderstand how they work and attribute human-like psychological traits to them. This trend is particularly problematic with social chatbots designed to become virtual companions. This technology is not necessarily harmless; but we can mitigate potential risks by communicating about how chatbots work, and ensuring that they are marketed and used responsibly."

A general audience talk on the recent progress of large models in AI and how they perform a form of artificial mimicry that has potentially concerning ethical implications.

An interview about my Nautilus article regarding the capacities, limitations, and potential risks of large pre-trained AI models.

A long-form conversation on the progress and future of artificial intelligence, spanning topics such as the definition of AI, the nature of intelligence, natural language understanding, out-of-distribution generalization, the benefits and limitations of scaling AI models, existential risk, and compositionality.

"What exactly is artificial intelligence? Will there ever come a time when artificial intelligence will become as competent as a human being? Our tenth episode features Dr. Raphaël Millière, who is the Robert A. Burt Presidential Scholar in Society and Neuroscience at Columbia University. In this episode, Saurish and Dr. Millière explore the philosophical questions that surround artificial intelligence, as well as ethical considerations."

A discussion about the natural language processing algorithm GPT-3, and its potential benefits and risks to society.

A public experiment on the relationship between taste and motion perception, organized with Barry C. Smith and Ophelia Deroy.