New Frontiers of AI: How Columbia Researchers Are Shaping Ethical and Impactful Innovation

The Columbia AI Summit Mar. 4 brought together leading experts to explore the latest breakthroughs in artificial intelligence. At the session, “New Frontiers of AI”, the speakers addressed both the speed of technical advancement and the pressing need for ethics, interpretability, and human-centered design.

“We're swimming towards one of the biggest waves in technologies that my generation has seen,” said Columbia University Provost Angela V. Olinto, also a professor of astronomy and physics, speaking to a packed audience at Low Memorial Library. “Columbia is a leader in this great rush to develop AI and ensure it is developed responsibly, for the benefit of all of us on this wonderful planet.”

Vishal Misra, vice dean for computing and AI and professor of computer science, and Richard Zemel, the Trianthe Dakolias Professor of Engineering and Applied Science in the Department of Computer Science and director of the NSF AI Institute for Artificial and Natural Intelligence (ARNI), moderated the event.

Image

Advancing AI toward human-like intelligence

A central theme of the session, which kicked off the day-long AI Summit, was the pursuit of AI that more closely mirrors human cognition. While AI systems have made significant strides in processing and generating language, they still fall short of true understanding.

Christos Papadimitriou, the Donovan Family Professor of Computer Science at Columbia Engineering, traced AI’s evolution from its inception in the 1950s to its explosive growth in the last decade. Despite the immense progress, he emphasized there are challenges such as AI’s opaque decision-making, its role in exacerbating inequality, and competitive pressures that compromise safety and ethics that must be addressed. He and researchers at ARNI are actively working on these issues, focusing on AI that better mimics human reasoning.

Lila Davachi, professor of psychology at the Graduate School of Arts & Sciences, examined the cognitive differences between humans and AI. Unlike AI, which has virtually unlimited computational resources, human learning is constrained by biological limitations, forcing efficiency. Understanding these differences, she argued, is essential for advancing both artificial and human intelligence.

Similarly, Elias Bareinboim, associate professor of computer science at Columbia Engineering, explored how embedding causal reasoning into AI can lead to fairer and more interpretable systems. Rather than relying solely on pattern recognition, AI needs to understand cause-and-effect relationships to make sound decisions. His lab is pioneering research in causal AI, which has the potential to revolutionize areas such as healthcare, policymaking, and criminal justice by reducing bias, improving medical treatment plans, and enhancing decision-making transparency.

Bridging the gap between AI and robotics

While AI has made significant progress in processing information, robotics faces unique challenges in interacting with the physical world. Yunzhu Li, assistant professor of computer science at Columbia Engineering, highlighted the need for better sensing, understanding of dynamics, and data collection in robotics.

Unlike AI language models, which operate in digital environments, robots must navigate and manipulate the real world. Despite advancements in self-driving technology and physics-based operations, robots still struggle with complex physical interactions. Li’s lab is developing advanced tactile sensing systems and multimodal learning techniques to enable robots to interact with their surroundings more effectively. His ultimate goal is to create robots that, like AI language models, may not always be perfect but are consistently capable of generating reasonable and useful actions.

The future of AI: ethical, transparent, and human-centered

Across AI and robotics, researchers are pushing for systems that are not only more intelligent but also more ethical, transparent, and aligned with human values. As AI continues to shape industries, it is crucial to address the challenges of explainability, fairness, and safety.

The future of AI and robotics is brimming with potential—but realizing that potential requires thoughtful innovation. Columbia’s researchers are ensuring that AI evolves into a tool that complements human intelligence rather than replacing or complicating it. Through ethical and impactful research, they are paving the way for a future where AI serves humanity in meaningful and responsible ways.

“AI is still far behind the human brain,” said Papadimitriou. “There is still a lot of work to do.”

Lead Photo Caption: Elias Bareinboim, associate professor of computer science at Columbia Engineering

Lead Photo Credit: Eileen Barroso/Columbia University