"Generative AI: With Great Power Comes Great Responsibility" - Prof. Ben Shneiderman скачать в хорошем качестве

"Generative AI: With Great Power Comes Great Responsibility" - Prof. Ben Shneiderman

Lecture by Professor Ben Shneiderman from The University of Maryland at UNSW, 24 March 2025. Generative AI Tools, like ChatGPT, Gemini, Copilot, Firefly, and Midjourney, provide powerful features for users, but also bring substantial dangers. Generating text, answering questions, summarizing documents, and writing code are common text-based applications, complemented by multimedia images, videos, songs, and speech. Generative AI products are startlingly impressive, but alarmingly flawed. Potential dangers include errors from “hallucinations”, biased statements, violations of privacy, and copyright infringement. Can these dangers be prevented or mitigated sufficiently to bring commercial success beyond the initial bubble of excitement? This talk will address these opportunities and dangers, with applications in education, medicine, and creative industries. We will focus on how HCI researchers and user experience designers can develop control panels that enable users to get what they want more easily and more often. The guiding Human-Centered AI principles stem from human moral and legal responsibility for the uses of technology. For researchers and developers, the message is: You are responsible for what you implement, so develop with guardrails and test repeatedly. For users, the message is: Generative AI is powerful, so learn to use it; Generative AI is flawed, so be careful. This event was co-hosted by the UNSW School of Computer Science and Engineering and UNSW AI Institute.