Annual Meeting 2024

Title:  Responsible AI in the Generative Era

Abstract:  Responsible AI (RAI) seeks basic guarantees of fairness, safety, privacy, robustness, controllability, explainability, transparency, and governance for traditional ML systems and generative AI systems. With recent legislation, including the EU AI Act, RAI has become a central focus of the AI/ML product development cycle. This talk provides an overview of the current practices for measuring and mitigating RAI dimensions in generative systems, and recent research in AWS AI/ML.

Vanessa Murdock leads a research group in AWS AI/ML, whose focus is Responsible AI (fairness, safety, privacy, robustness, and veracity).  In addition to doing fundamental research in RAI topics, her team builds tools for assessing aspects of responsible AI, used in AWS Services such as Bedrock, Rekognition and Transcribe.  Prior to joining AWS, she led a science team in Alexa Shopping focused on recommender systems, search and HCI. Her team provided the machine learning that backed Amazon’s Choice, and Alexa Shopping List, in addition to contributing content moderation for the generative AI system Rufus. She was previously at Microsoft, working on location inference and notifications at Bing and Cortana. Prior to Microsoft, Murdock led the Geographic Context and Experience Group at Yahoo! Research in Barcelona, which centered on geographic information retrieval and user-generated content. She has been awarded 20 patents, resulting in a Master Inventor Award from Yahoo! (2012). She received the OAA Award for Outstanding Achievement by a Young Alum from the University of Massachusetts in 2014. She is currently serving as the Chair of the ACM SIGIR Executive Committee.  Murdock received a Ph.D. in Computer Science from the University of Massachusetts Amherst Center for Intelligent Information Retrieval, advised by Bruce Croft.

Kayleen Devlin is a senior journalist at BBC Verify. She has extensive experience of working on open-source investigations and covering disinformation from the Ukraine and Israel Gaza wars. As well as this, she has also covered topics ranging from climate denial around COP26 and election related disinformation in the Philippines and the US midterm and presidential elections.

In a year filled with global conflicts and elections, communicating credible information in a timely manner has never felt more important. BBC Verify senior journalist, Kayleen Devlin, joins us to discuss some of the approaches her team takes when it comes to tackling misleading posts online. What are some of the main themes that have cropped up this year? And how much has generative AI disrupted the landscape?

Sanja is an applied mathematician with MSc in cybersecurity and cryptography and a PhD in data science. Almuna of Internation Space University and ex of ICEYE, she has a keen interest in space technology and research (e.g., AI for Earth Observation).

Her professional experiences include government institutions, two startups, CERN, and Bell Labs. During her doctoral studies with EIT Digital, she has also taken business, innovation and entrepreneurship courses, working on startup ideas in some of them.

Her example research projects include public and population health studies using social networks, human dynamics using mobile phone data, and urban vitality using satellite data.

Abstract: This talk presents the work of the Responsible AI (RAI) team at Nokia Bell Labs, Cambridge. It covers the six RAI pillars and solutions for designing, deploying, and monitoring responsible AI systems. Discover AI Design, a collaborative approach for holistic AI system design, and ExploreGen, which helps foresee and manage potential uses and risks of AI technology. See how NLPGuard prevents over-reliance on protected attributes when monitoring a toxicity classifier. Finally, gain insights into how AI incidents are portrayed in the news media.

Siri Gloppen is a Professor of Political Science and Vice Dean at the Faculty of Social Sciences. She directs LawTransform, the CMI-UiB Centre on Law & Social Transformation, and heads the Bergen School of Global Studies. Her work focuses on the intersections of law, politics, and social change, and she is actively involved in global development and policy research initiatives.

Birkeland holds a Master of Science in Engineering from NTNU in Trondheim and has a background as CEO of RiksTV.

Christian Birkeland is part of the executive management team at TV 2 and the chairman of the steering board in SFI MediaFutures.

Svela is an award winning investigative journalist and editor. As a journalist in Bergens Tidende, Svela won a Skup-diploma for the fact checking initiative faktasjekk.no. Svela held various leadership positions over 11 years in Bergens Tidende, og was in charge of the newspaper’s coverage of the 2011 terrorist attack. Svela has also led various innovation initiatives in Bergen Tidende, among them New digital formats, which won a silver medal in the 2022 INMA Awards for innovation in newsroom. He started as the CEO of Media Cluster Norway in September 2022. Media Cluster Norway is a media and media tech cluster with around 90 member companies. Svela chairs the ITPC working group on Provenance Best Practices and Implementation. 

Chris Ronald Hermansen is the project manager for editorial AI at TV 2. Over the past 15 years, he has worked as a journalist, news editor, and project manager in various media companies. In recent years, he served as the editorial manager for TV 2’s news department in Bergen. Hermansen holds a law degree.

Lubos Steskal is a data scientist and AI developer at TV 2’s editorial AI team, and he is also the industry lead of the Media Futures language technology work package. Combining academic and industry experience, he has worked extensively with natural language processing and machine learning across various sectors. His background includes positions at the University of Bergen, Sbanken, and several startups, bringing a unique perspective to the intersection of AI, journalism and media.

‘KI-Kjetil’, the first AI avatar of a news personality in Norwegian media, serves as an interactive chatbot focusing on U.S. presidential election coverage. Based on journalist and news anchor Kjetil H. Dale, this talk presents the journey of developing and deploying KI-Kjetil, from evaluating editorial questions and challenges through technical implementation and towards operational monitoring. We explore the editorial rationale behind creating an AI clone of a news anchor, discuss our project objectives, and evaluate their outcomes. The presentation opens a discussion about the broader implications of AI avatars in journalism, examining their impact on public trust and the ethical dimensions of deploying AI-powered representations of real journalists.

Sergej  is the Chief Innovation Officer at Wolftech, where he leads a culture of innovation across interdisciplinary teams, transforming industry standards for a platform serving over 20,000 global professionals. With a PhD in Computer Science and recipient of the Eurovis Best Dissertation Award, Sergej combines research-driven strategic planning with a passion for customer-centric solutions. He is a recognized AI expert and thought leader, regularly speaking at key industry events and driving sustainable growth through strategic partnerships and groundbreaking AI solutions for media professionals.

Morten Langfeldt Dahlback leads the technology and development efforts at Faktisk.no. He also heads the EU project NORDIS, a Nordic collaboration aimed at countering misinformation and disinformation. Dahlback holds a PhD in philosophy and has previously worked as a commentator for Adresseavisen and as an analyst for The Economist Intelligence Unit.