ADAPT Radio The ADAPT Centre

AI is developing at such a rapid pace that we can get caught up in its potential capabilities and role in our future. However, there are still a lot of issues to rule out.

ADAPT recently hosted the Annual Scientific Conference 2024 in Dublin and today we’re hearing one of the keynote speakers, Abeba Birhane. We learn about the potential dangers of large-scale datasets, such as AI hallucinations and the reinforcement of societal biases and negative stereotypes. She also explored strategies for both incremental improvements and guiding broader structural changes in AI.

Our expert guest has been exploring strategies for both incremental improvements and guiding broader structural changes in AI. She is Senior Advisor for AI Accountability at Mozilla, Adjunct Professor at Trinity College Dublin and new ADAPT member, Abebe Birhane.

THINGS WE SPOKE ABOUT

● How rumours of autonomous AI distract from real issues
● Hallucinations creating factually incorrect information
● AI ownership giving power to the hands of the few
● Data issues with collection, copyright and biases
● Creating standards for the safe use and development of AI

GUEST DETAILS

Abeba Birhane is a cognitive scientist, currently a Senior Advisor in AI Accountability at the Mozilla Foundation and an Adjunct Assistant Professor in the School of Computer Science and Statistics at Trinity College Dublin (working with Trinity’s Complex Software Lab).

She researches human behaviour, social systems, and responsible and ethical artificial intelligence and was recently appointed to the UN’s Advisory Body on AI. Abeba works at the intersection of complex adaptive systems, machine learning, algorithmic bias, and critical race studies. In her present work, Abeba examines the challenges and pitfalls of computational models and datasets from a conceptual, empirical, and critical perspective.

Abeba Birhane has a PhD in cognitive science at the School of Computer Science, UCD, and Lero, The Irish Software Research Centre. Her interdisciplinary research focused on the dynamic and reciprocal relationship between ubiquitous technologies, personhood, and society. Specifically, she explored how ubiquitous technologies constitute and shape what it means to be a person through the lenses of embodied cognitive science, complexity science, and critical data studies.

Her work with Vinay Prabhu uncovered that large-scale image datasets commonly used to develop AI systems, including ImageNet and 80 Million Tiny Images, carried racist and misogynistic labels and offensive images. She has been recognised by VentureBeat as a top innovator in computer vision.

MORE INFORMATION

Adapt Radio is produced by DustPod.io for the Adapt Centre
For more information about ADAPT visit www.adaptcentre.ie/

QUOTES

Generative AI has been around for quite some time, but the introduction of DALL-E back in April 2022 can be noted as one of the landmarks where generative AI really exploded into the public space. - Abeba Birhane

These hypothetical AI concerns, existential concerns, about the very idea of this attempt to build AGI has neither scientific nor engineering principles. Again, a lot of it is just hype, marketing, and PR, that is just really dominating the entire field. - Abeba Birhane

The results are really worrying, over 50% of the output from these models was inaccurate, 40%, harmful, and incomplete - Abeba Birhane

There is no such thing as fully autonomous AI, we will always need people, humans, in the loop. - Abeba Birhane

KEYWORDS
#ai #data #audit #research #chatgpt #ethicalai

How we view the past is very much shaped by the stories we're told about it and AI innovations are helping researchers to enrich our understanding of history.

A new project called VOICES is using artificial intelligence to uncover the untold women's experiences of extreme trauma and civil war in early modern Ireland. The project was recently launched at Trinity College Dublin and today we hear from the interdisciplinary team of historians and computer scientists leading the research.

Our expert guests are two Trinity researchers, Erasmus Smith's Professor of Modern History, Jane Ohlmeyer, and Professor in Computer Science in the School of Computer Science and Statistics, Declan O'Sullivan.

THINGS WE SPOKE ABOUT

● Challenges in processing data from the 16th and 17th Century
● Utilising knowledge graphs for interpreting phrases, spelling and grammar
● Recovering stories of Ireland’s forgotten women
● Creating processes and digital archives for future researchers
● Crossovers in learning from interdisciplinary approaches

GUEST DETAILS

Professor Jane Ohlmeyer is Erasmus Smith's Professor of Modern History at Trinity College Dublin and Chair of the Irish Research Council, which funds frontier research across all disciplines. Jane was the founding Head of the School of Histories and Humanities, Trinity’s first Vice-President for Global Relations (2011-14) and Director of the Trinity Long Room Hub Arts and Humanities Research Institute (2015-20).

Jane is an expert on the New British and Atlantic Histories and has published extensively on early modern Irish and British history. She is the author or editor of numerous articles and 13 books. Jane has been the Principal Investigator (PI) or co-PI for 25 research and research infrastructure projects with awards totalling c.€22 million from national, European and international funders.

Declan O’Sullivan is a Professor in Computer Science at the School of Computer Science and Statistics, and is a co-applicant Principal Investigator in the ADAPT SFI Research Centre. Prof. O’Sullivan and his team’s research in Knowledge Graph Techniques, which extracts, transforms and integrates data, is central to this relationship between data and machine learning.

Since joining TCD from industry in 2001, Declan has established himself as an international research leader in his field: authoring 260+ scientific peer-reviewed papers and international Journals; being a member of 3 journal editorial boards and having undertaken 12+ chair roles in IEEE and IFIP conferences over the years. He has won competitive research funding as PI and Co-PI of approximately 7.8M euro. He was elected as a Fellow in Trinity College Dublin in 2019 in recognition for the quality of his contributions.

MORE INFORMATION
Adapt Radio is produced by DustPod.io for the Adapt Centre
For more information about ADAPT visit www.adaptcentre.ie/

KEYWORDS
#knowledgegraph #ireland #data #history #17thcentury #archives #AI

You cannot move an inch these days without encountering takes on the future of AI and technology, and concerns about how it may impact our lives for better or worse.

Today we're talking about new ways to approach the ethics of AI and digital technologies and what research is being done to answer the questions and the dilemmas these new technologies raise.

Our experts today are from the ADAPT Centre and are researching alternative methods of ethical and critical thinking in the design of digital technologies and AI.

They are Assistant Professor at the School of Information and Communication Studies in University College Dublin (UCD), Dr Marguerite Barry and postdoctoral researcher at the School of Information and Communication Studies at UCD, Dr Paul O’Neill.

THINGS WE SPOKE ABOUT

● Critical thinking and ethical design in AI
● Incorporating ethics at the research stage
● The importance of public engagement on future uses of digital technologies
● Beta Festival: Using art to encourage engagement and critical thinking
● Interdisciplinary, multidisciplinary and transdisciplinary research

GUEST DETAILS

Prof Marguerite Barry is Assistant Professor at the School of Information and Communication Studies in University College Dublin (UCD). Her research area is human-computer interaction (HCI) and digital media communication studies with a focus on ethical design and development in policy and practice. She is a funded investigator with ADAPT on the Transparent Data Governance strand where she is working on the Autonomy & Responsibility challenge. This involves interdisciplinary projects to support multi-stakeholder engagement in AI technologies from design to deployment.

Dr. Paul O’ Neill is an artist and researcher whose practice and research are concerned with the implications of our collective dependency on networked technologies and infrastructures. Paul is a postdoctoral research fellow at the ADAPT Centre for AI-driven Media Technologies at University College Dublin where he is focusing on the ethics and design of Artificial Intelligence systems. He is also a co-curator of the Dublin Art and Technology Association (D.A.T.A).

MORE INFORMATION

Adapt Radio is produced by DustPod.io for the Adapt Centre
For more information about ADAPT visit www.adaptcentre.ie/

KEYWORDS
#technology #ethics #data #research #ai #art

The Irish language is an important part of Ireland's culture, but it is a minority language and like many ‘at risk’ languages around the world, Irish needs to be protected.

Today we're talking about how AI can help to boost the Irish language and the importance of diverse data collection in building robust translation systems. We also hear how researchers are using natural language processing and other tools to help maintain the richness of the language and make it more accessible and available to those who use it.

Our experts today are passionate about protecting minority languages and are working on technology to improve machine translation of the Irish language with the Adapt Centre. They are postdoctoral researcher, Dr Abigail Walsh and research assistant with eSTÓR, Gráinne Caulfield, both from Dublin City University.

THINGS WE SPOKE ABOUT

● Protecting minority languages with AI
● The limitations of data collection and the need for more diversity
● Using Natural Language Processing to collect the complexities of a language
● AI’s role in encouraging more use of an at risk language
● Getting social media platforms on board to make language more accessible

GUEST DETAILS

Abigail Walsh is a PhD student at the ADAPT Centre in Dublin City University. Her research focuses on improving NLP for Irish language, focusing on the treatment and automatic processing of Multiword Expressions (MWEs). Abigail’s interests include Irish language technology, MWEs, NLP for low-resource languages, linguistic analysis, data processing, Machine Translation, and Machine Learning.

Gráinne Caulfield is a recent graduate of Irish & French from Trinity College Dublin, currently working as a Research Assistant on the eSTÓR project. In this role, she executes the project’s outreach activities- carrying out site visits to relevant stakeholders, managing the social media platforms, newsletter writing etc, as well as translation and data processing duties.

MORE INFORMATION

Adapt Radio is produced by DustPod.io for the Adapt Centre
For more information about ADAPT visit www.adaptcentre.ie/

KEYWORDS
#irish #language #data #translation #machinetranslation #technology #ai

The pace of change in AI is beyond rapid and it's an exciting time for research in Ireland.

Professor John Keller is now leading the way in research as the new director of ADAPT, the SFI Research Center for AI Driven Digital Content Technology.

Today we hear about John’s ambitions in his new appointment including his thoughts on current developments in AI, how he plans to lead ADAPT and prepare for the future and why supporting multidisciplinary research will be a key focus for the centre.

THINGS WE SPOKE ABOUT

● Keeping up with the pace of AI as it enters the mainstream
● SignON: Making AI accessible
● ADAPT’s role in mitigating the harms of AI
● Integrating multidisciplinary and interdisciplinary research
● Mentoring and resourcing new researchers and culture
● Preparing for future challenges across different industries

GUEST DETAILS

Prof John D. Kelleher is the newly appointed director of the ADAPT Centre and Chair of Artificial Intelligence at Trinity College, Dublin. John’s core research expertise is in the areas machine/deep learning and natural language processing. Previously, he was the TU Dublin lead in the ADAPT centre and the scientific lead for the Digital Content Transformation Strand.

Within the ADAPT Centre he leads research projects on language modelling, lexical semantics, machine translation, novelty detection, image captioning, dialog systems, and making AI more environmentally sustainable.

John has been the academic lead on numerous industry projects across a range of topics and domains, including: anomaly detection, transfer learning, customer segmentation and propensity modelling, dialog systems and chat bots, and information retrieval and natural language processing.

MORE INFORMATION

Adapt Radio is produced by DustPod.io for the Adapt Centre
For more information about ADAPT visit www.adaptcentre.ie/

QUOTES

The impact and penetration of AI into people's lives means that there's so much interest from industry in artificial intelligence, and that's really driving the pace. - John Kelleher

When these types of technologies are prevalent throughout society it's important that the people in that society understand what they are and how they work. - John Kelleher

The most important aspect, or the foundations we need to build in order for interdisciplinary research to flourish, is that we can communicate across disciplines. So we need to spend time with each other as researchers, understanding each other's language. And at the core of that is a willingness to listen. - John Kelleher

Exciting things often happen at the margins, between disciplines, for people to come together in a complementary team, where they have complementary expertise. And so once you can build a bridge across the discipline, then you can do really exciting work. - John Kelleher

Fluency in and of itself isn't the sign of intelligence that we thought it was before. And that's why maybe we need more emphasis on creativity and critical thinking within education. - John Kelleher

KEYWORDS
#ai #adapt #research #technology #artificialintelligence #culture #languagemodels #accessibility

In the age of data our private information is currency and we digital trails behind us everywhere we go online. As AI grows in popularity, some fear it may be a threat to our privacy. It’s important we consider how we can best protect ourselves from our valuable information ending up in the wrong hands.

Today, in light of Data Privacy Week, we're diving into the week’s theme, ‘Take Control of Your Data’, to ask how we can do that, who the responsibility of protection and regulation lies with and the ways generative AI can be a game changer for data privacy.

Our expert is passionate about data protection at an organisational level and is Vice President and Chief Knowledge Officer at the International Association of Privacy Professionals, Caitlin Fennessey.

THINGS WE SPOKE ABOUT

● How the International Association of Privacy Professionals (IAPP) is approaching AI
● Generative AI and protecting data privacy
● Data Privacy Week: Taking control of your data
● Encouraging organisations to embrace AI, rather than ban it
● Global streamlining of data privacy regulation
● Individual responsibility and creating change in organisations

GUEST DETAILS

Caitlin Fennessy is Vice President and Chief Knowledge Officer at the International Association of Privacy Professionals, where she guides the strategic development of IAPP research, publications, communications, programming and external affairs.

Caitlin is a recognized privacy expert, serving as an inaugural member of the UK International Data Transfers Expert Council, on the German Marshall Global Task Force to Promote Trusted Sharing of Data and on the Future of Privacy Forum Advisory Board. She speaks and leads frequent public discussions on the practical impacts of privacy developments around the world.

Prior to joining the IAPP, Caitlin was the Privacy Shield Director at the U.S. International Trade Administration, where she spent ten years working on international privacy and cross-border data flow policy issues.

International Association of Privacy Professionals: https://iapp.org/about/person/0011a00000DlNmBAAV/

MORE INFORMATION

This episode is in association with Empower, which is coordinated by Maynooth University.

For more information about ADAPT visit www.adaptcentre.ie/

Adapt Radio is produced by DustPod.io for the Adapt Centre

KEYWORDS
#privacy #data #ai #governance #privacyissues #artificialintelligence