Director's Pick: AI and the Future of Intelligence скачать в хорошем качестве

Director's Pick: AI and the Future of Intelligence

March 18, 2025 Director's Pick AI and the Future of Intelligence In this lecture, Professor James Evans of the University of Chicago will introduce the power of "large models" (LMs) underlying modern artificial intelligence (AI) as arbitrary function approximators that predictively simulate text, sound, video, proteins, action sequences and more. As most LM data today was generated by humans, first summoned by the public Internet, LMs are cultural technologies, like language, libraries, and bureacracies, which compress, recombine and communicate information in new ways. Generative AI can also act. As agents assisting humans in the world, they now manifest human-like intelligence, biases and their collective consequences, ranging from elaborate deceptions to ideological polarization. Nevertheless, LMs reproduce the complex world we feed them with alien architectures they have increasingly invented themselves, and which scientists now interpret to understand AI foresight and the emerging principles underlying it. Surprising properties of emergent AI intelligence make their thinking and behavior complementary to achieving human wants and needs. Increasing engagement with and exchange between humans and AI portends a kind of "major transition" seen in the emergence of biological complexity, where we incorporate LMs and they incorporate us to alter and enhance collective intelligence. In these evolving human-AI systems, Professor Evans will explore challenges associated with producing as well as consuming diversity to achieve sustained innovation and avoid system collapse, for example, demonstrating how scientists have succeeded in their use of AI to think and know more, but their pathways toward established data are correlated and so AI-infused science, up until now, has focused attention on a narrower swath of known problems rather than opening new concerns with new sensors and data. Professor Evans will conclude with a discussion of existential concerns around AI safety for humanity, and efforts to ensure it that range from disciplining AI to slavishly honor human values to building a diverse ecology of AIs that check and balance themselves. Speaker: James A. Evans Max Palevsky Professor of Sociology and Data Science Director of Knowledge Lab Founding Faculty Co-Director of the Computational Social Science Program The University of Chicago Moderator Mark Barnekow (MBA’88) Executive Director The Hong Kong Jockey Club University of Chicago Academic Complex | The University of Chicago Francis and Rose Yuen Campus in Hong Kong Read more about Director’s Pick: https://bit.ly/3NZEEQK ➡ Subscribe to our Enews: https://bit.ly/3fxp8JP Twitter:   / uchicagohk   Facebook:   / uchicagohk   Linkedin:   / uchicagohk   Instagram:   / uchicagohk   ▬ Content of the Webinar ▬ Opening - 0:00 Speakers Introduction - 00:54 Speakers' Sharing - 02:18 Q&A session - 56:27