Access the event on demand here.
The third annual ADVANCE event, organised by the ADAPT Centre, took place on Wednesday, June 7th, bringing together renowned academics and industry experts to discuss the advancements and future prospects of artificial intelligence (AI). This half-day conference focused on the theme “Generative AI, Large Language Models and Beyond: Strategies for Adoption and Integration.” Attendees had the opportunity to engage with the latest research developments and gain insights into the potential applications of large language models across industries. Here, we summarise the key contributors and highlights from the event.
The event began with a welcome and introduction to ADAPT by Professor Dave Lewis, Director of the ADAPT Centre. Professor Lewis highlighted ADAPT’s belief in opting for a multi-disciplinary approach to science and technology that includes experts from different disciplines such as sociology, ethics, law, clinician and engineering in creating the future we want rather than a future that we may not all benefit from.
Professor Anya Belz, a prominent figure in Natural Language Generation and evaluation of Natural Language Processing (NLP) systems, showcased her research in machine learning and NLP. During her talk, Professor Belz gave a first look at large pre-trained language models, the latest generation of neural language models that have surprised everyone with the astonishing variety of things they can do and the quality with which they can do them. In her talk she looked at where their power comes from, what they can currently do, and what new capabilities might emerge in the future. The amount of time, money and resources it takes to train a base model was highlighted along with some example application tasks that PLMs can achieve state-of-art performance on straight out of the box. Professor Belz finished her talk on why we all need training in prompt engineering now, and where tool-augmented language models might take us in the future.
A second expert talk was provided by Professor John D. Kelleher, a renowned expert in machine learning and natural language processing. Professor Kelleher discussed the opportunities, challenges and future direction for LLMs and generative AI. He emphasised the importance of making AI more environmentally sustainable and outlined ongoing industry projects related to natural language processing.
The day also included two panel discussions featuring leaders in a range of industries. Both panels were chaired by Jonathan McCrea, a prominent science and technology broadcaster and founder of Whipsmart Media. Jonathan explored various aspects of living and working with large language models. The discussions delved into understanding the impact of these models on our world, exploring their potential applications, developing ethical guidelines, and discussing best practices for their effective use. The first panel was titled, ‘Generative AI: A disruptive impact on industry?’ and featured Dr Robert Ross of the ADAPT Centre at TU Dublin, Lúí Smyth the Director of Product, Generative & AI Tooling at Shutterstock, and Dr Mary Coughlan, Partner in EY and the Irish firm’s leader for Health Data and Analytics. The second panel was titled ‘Data, Ethics and AI Regulation: A New World of Generative AI’ and featured Dr Patricia Scanlon, Ireland’s first AI Ambassador and founder of SoapBox Lab; Senator Malcolm Byrne, Fianna Fáil’s Spokesperson on Further and Higher Education, Research, Innovation and Science; Aoife Sexton, Chief Privacy Officer and Chief of Product Innovation in Truata; and Professor Dave Lewis, Director of ADAPT.
The event also featured industry insights from professionals at the forefront of AI development. Paul Barnes, a Senior Program Engineer at Microsoft, discussed the role of large language models in enhancing productivity tools such as Microsoft Editor. He emphasised the power of writing intelligence and its potential to transform work experiences. Dr Cian Hughes, who works at Google Health. Cian has worked on innovations using machine learning in diagnostic settings. His recent research on LLMs in the medical domain as part of Google’s Med-PaLM work highlighted the potential for the technology for clinicians.
A full recording of the event is available online.