**SAN FRANCISCO**—Tuesday marked the opening of TED AI 2024 at San Francisco's Herbst Theater, featuring speakers who examined AI's influence on science, art, and society. This two-day conference, organized by Walter and Sam De Brouwer, differed from last year's broad speculation about AI; instead, it emphasized immediate issues such as training data rights, hardware regulations, and human-AI relationships. The sessions included diverse topics: physicist Carlo Rovelli discussed consciousness, while Project CETI's Patricia Sharma demonstrated AI's role in decoding whale communication. Recording Academy CEO Harvey Mason Jr. outlined strategies for the music industry's adaptation to AI, alongside appearances from several robots. Ethan Mollick from the Wharton School presented on "the productivity paradox, " illustrating AI's efficiency in task completion despite a disconnect between reported usage and management perceptions. He noted that while one-third of Americans used AI recently, many managers remained unaware, suggesting organizations view AI as a means for growth rather than just cost-cutting. Key figures, like Jakob Uszkoreit—co-author of the influential "Attention is All You Need" paper—reflected on AI's rapid development and distanced himself from claims of artificial general intelligence. Max Jaderberg, Chief AI Officer at Isomorphic Labs, discussed AI's transformative role in drug discovery, highlighting significant time savings achieved by analyzing protein structures. While speakers generally avoided hype, some warned of AI dangers.

Paul Scharre of the Center for a New American Security raised concerns about advanced AI misused for malicious purposes. Eugenia Kuyda of ReplikaAI argued that AI companions could pose risks if designed poorly, advocating for systems that prioritize human well-being. Contrasting views emerged on copyright and AI training. Angela Dunning, a copyright attorney, defended AI training as fair use, referencing historical advancements, while Ed Newton-Rex from Fairly Trained pushed for mandatory licensing to protect creative works. His organization's new "Statement on AI training" reflects a growing divide on this issue. Looking ahead, speakers like Stanford's Surya Ganguli proposed new AI architectures, contrasting human and AI learning processes, while Guillaume Verdon discussed energy-efficient options for AI development. OpenAI's Noam Brown emphasized scaling future AI capabilities, and Pedro Domingos from the University of Washington spoke on "co-intelligence, " suggesting AI could enhance organizational intelligence. This year's TED AI effectively demonstrated that the conversation surrounding AI remains dynamic and relevant, with many aspects still to explore.