Ikigai’s Post

When talking about AI fairness and ethics, the debate often tends to focus on the abstract principles but there are some very practical considerations that make LLMs more or less viable for specific use cases in research. For example we've been exploring ways of using LLMs in Public Opinion Research and this paper from Sungkyunkwan University and Hamad Bin Khalifa University is super interesting. The study shows how if you look at LLMs outputs, they tend to match public opinion from polling when analysing audiences in aggregate. But run an analysis of any subgroup and you realise that the LLM outputs start degrading away from the polled outputs, particularly when the subgroup is defined by race, gender, politics and the economy. Granted this was tested on GPT 3.5 Turbo so things might have changed with recent upgrades, but the point is yes you can glean insights on public opinion from an LLM in aggregate but the moment you start looking at subgroups, about 60% of your results are gonna skew away from reality due to multiple forms of bias in the data. Full Study here https://lnkd.in/eXaZq3AG

Attended an insightful keynote by Dr. Michael Zimmer from Marquette University on data ethics and a new approach to AI ethics through discernment. Dr. Zimmer highlighted several conceptual gaps in data collection practices, urging us to rethink privacy, consent, and harm in digital spaces. We need to reconsider what it means to be a human subject in this context. For instance, just because data is public, does it mean researchers can freely use it? And does clicking "consent" upon creating an account grant platforms unrestricted use of that data? As an ethnographer, I'm particularly concerned about reflexivity and the potential impacts on my subjects. Dr. Zimmer emphasized that data scientists and computer scientists often overlook these ethical considerations. He argued that these professionals could learn from anthropologists, who are consistently aware and reflective of the power dynamics in their engagements with spaces, communities, and subjects. This lecture reinforced that ethics goes beyond simply checking boxes. Public data isn't a free-for-all. As researchers, we have a duty to protect our subjects and the spaces we study. SPATIAL Project #DataEthics #AI #Privacy #Consent #ResearchResponsibilities #Ethnography

Excited to highlight the "Responsible AI Development" panel at the finale of the Responsible AI in Global Business Conference on March 1st, 2024. Join us to delve into the evolving frontier of AI regulation, explore the implications of the EU AI Act, and uncover the complex issues faced by practitioners daily, with highlights of the pressing ethical challenges introduced by AI in the coming years. This panel features the top voices in the field: - Sasha Brown, Head of Ethics Foresight at Google DeepMind - Google’s world leading AI research lab. She works to ensure that Google DeepMind’s research, and the applications it powers, have a positive impact on humanity. - Luciano Floridi, Founding Director of the Digital Ethics Center (DEC), Yale University and Professor in the Practice, Cognitive Science Program at Yale University. This panel will be moderated by Edward Wittenstein, Director of the Schmidt Program on Artificial Intelligence, Emerging Technologies, and National Power at the Yale Jackson School of Global Affairs. Register here: https://lnkd.in/e7mn7-gn #ResponsibleAI #AIGovernance #AIPolicy #AIRegulations #AIInnovation #DigitalEthics #AIResponsibleInnovation #AIforGood #TechPolicy #AIResearch #YaleEthics

🌟 What our professors say about the Responsible AI Conference?🌟 The conference's realization was greatly supported by distinguished professors and leaders throughout the Yale community. In today's update, we spotlight the insights of Professor Luciano Floridi. Luciano Floridi, Founding Director of the Digital Ethics Center (DEC), Yale University Center and Professor in the Practice in the Cognitive Science Program at Yale University, brings unparalleled expertise in digital ethics and AI philosophy. His extensive contributions to the field, including over 300 publications and groundbreaking books, make him a key voice in shaping responsible AI practices globally. "We speak a lot about the future of AI, machine learning in particular, large language models. But we need to understand that it is through regulations, good policy, good strategy that all the benefits of AI will become a reality. Policy is not just about federal law or European law or international agreements, but it's also what we do, for example, at the university, what we do within a particular business. It is a matter of governance of technological innovation." — Professor Luciano Floridi Luciano’s full interview: https://shorturl.at/wCZ24 Stay tuned for more inspiring highlights: https://lnkd.in/eSqiwCfY Presented by the Yale SOM AI Group and International Center for Finance at Yale School of Management, this conference aims to foster essential dialogues on responsible AI adoption, development, and governance. In collaboration with esteemed partners such as the Yale Program on Stakeholder Innovation and Management, the Schmidt Program at Yale Jackson School of Global Affairs Affairs, and the Digital Ethics Center (DEC), Yale University, we are committed to advancing ethical practices in AI. #AIethics #ResponsibleTech #DataPrivacy #AItransparency #TechEthics #ResponsibleInnovation #AIgovernance #DigitalPolicy #FutureofAI #AIConference #YaleEthics #YaleUniversity #YaleLeadership #YaleResearch #YaleCommunity #CognitiveScience

In the current environment characterized by the rapid advancement and widespread adoption of AI technologies, there has been a proliferation of ethical guidelines formulated and disseminated by various organizations. The centrality of ethics in informing AI development is underscored in International Observatory on the Societal Impacts of AI and Digital Technology (OBVIA)'s report "Ethics at the Heart of AI." It offers five principal recommendations: 1. Offer to actors in the science and technology sectors training activities on AI and ethics that include a perspective from the humanities and social sciences. 2. Support the development of an ethical competency among AI researchers and developers, law makers, and other actors involved in the development, marketing, use and regulation of AI. 3. Through conferences, workshops and other activities opened to a large audience, increase the awareness of citizens with regard to the various social impacts and ethical challenges related to the development and deployment of AI. 4. Through a participative and interdisciplinary perspective that involves citizens as well as civil society organisations, create spaces, instances and other mechanisms that promote reflection, discussion and evaluation of AI technologies and their use. 5. Put in place in universities and public organisations interdisciplinary committees specific to AI and computer or robotic technologies. https://lnkd.in/eyqXU3Sb International Observatory on the Societal Impacts of AI and Digital Technology (OBVIA), Geneva Graduate Institute, Jérôme Duberry, Lyse Langlois, #AIethics #TechEthics #FutureOfAI

We are thrilled to announce the "Responsible AI Development" panel concluding our upcoming Responsible AI in Global Business Conference on March 1st, 2024. This panel will feature distinguished experts in the field, offering profound insights into the principles, practices, current state, and future prospects of responsible AI development. Join us as we explore the critical dimensions of creating responsible AI technologies: Sasha Brown, Head of Ethics Foresight team at Google DeepMind - Google’s world leading AI research lab. She works to ensure that Google DeepMind’s research, and the applications it powers, have a positive impact on humanity. Luciano Floridi, Founding Director of the Digital Ethics Center (DEC), Yale University and Professor in the Practice, Cognitive Science Program at Yale University. Luciano's expertise spans digital ethics and the ethical implications of emerging technologies, providing invaluable perspectives on AI governance. This panel will be moderated by Ted, Edward Wittenstein, Director of the Schmidt Program on Artificial Intelligence, Emerging Technologies, and National Power at the Yale Jackson School of Global Affairs. Register now to engage in shaping the future of responsible AI: https://lnkd.in/eSqiwCfY Presented by the Yale SOM AI Group and International Center for Finance at Yale School of Management, this conference aims to foster essential dialogues on responsible AI adoption, development, and governance. In collaboration with esteemed partners such as the Yale Program on Stakeholder Innovation and Management, the Schmidt Program at Yale Jackson School of Global Affairs, and the Digital Ethics Center (DEC), Yale University, we are committed to advancing ethical practices in AI. #ResponsibleAI #AIGovernance #AIPolicy #EthicalAI #AIRegulations #AIInnovation #DigitalEthics #AIResponsibleInnovation #AIforGood #TechPolicy #AIResearch #YaleEthics

Tune into The Institute for Learning-enabled Optimization at Scale (TILOS)'s AI Ethics webinar on Friday, 3/8 at 12:00 p.m. PST! Attendees will learn frameworks for integrating ethics into AI development, strategies for measuring bias and fairness, and gain insights into how optimization can reveal and mitigate societal biases in AI algorithms. #aiethics #artificialintelligence #trustworthyai #optimization #biasinai #fairness

Sure, anyone can deliver on the technical rigor, but at the University of St. Thomas, we are also driven to put technology into context. The AI Ethics course in the University of St. Thomas School of Engineering AI program is an important lens into how AI can impactfully advance the work of our partner companies. And, it is just a part of reason we say we are building something truly special here in Minnesota. #ai #engineeringeducation #ethics

Over the past few weeks I've enjoyed the challenge of achieving my 'Ethics in AI' certification from The London School of Economics and Political Science (LSE). With a huge thanks to the extremely diverse range of very active participants and to Dr Cat Wade, PhD, Harvard University for her tutorship on such a challenging topic - my personal take aways are: - Ethics is hard! As someone who's used to numbers and a 'right'(ish) answer, it's been great to be stretched into the realm of value trade-offs, political systems and scrutinising my own beliefs. - Whilst at a personal level we may be happy to share all of our own information, as AI matures we risk exposing other individuals who want to remain private. What responsibility do we have to seriously consider what we share, in order to protect others? - (thanks to one of my classmates who raised this one!) Communicating with AI is like communicating with a human, in the sense that 'thoughts' are hidden and we don't actually know if we're being lied to. Where this differs is that (currently) we are able to build relationships and develop trust with other humans. How can we build relationships and trust with this new community of 'AI people'? #AI #Ethics #GotMeThinking #Data #Privacy https://lnkd.in/e8km52VW

"AI Ethics is a core component of our new Master of Science in Artificial Intelligence, essential for shaping responsible AI practitioners. As artificial intelligence rapidly transforms our world, emphasizing ethical considerations alongside technical skills, we prepare our graduates to lead in creating AI systems that are not only powerful, but also fair, transparent, and beneficial to humanity. " - Dr. Manjeet Rege, University of St. Thomas AI professor and Director of the Center for Applied Artificial Intelligence. Learn more about this important topic and the AI & Ethics graduate course co-taught by St. Thomas instructors Brandan Keaveny, Ed.D., CIPM and Jessi Benzel: https://lnkd.in/eUbYUJQn #ai #ethics #careers