Data Science Institutes from Imperial and LSE, co-hosted a symposium exploring how large language models like ChatGPT could transform society.
The Symposium, which took place on 24/25 May 2023, was the latest event to emerge from an ongoing partnership that exists between the sister institutes known as DSI Squared.
Over two days at Imperial College London and The London School of Economics and Political Science, the two Data Science Institutes (DSI) explored the technical basis, future directions, industry applications, and consequences of generative AI, with particular attention to the knowledge economy and intellectual workers.
Day One closed with a Keynote discussion with OpenAI’s Co-Founder Wojciech Zaremba, who was joined by OpenAI’s Product Lead Arka Dhar, and Imperial’s Vice Provost for Research and Enterprise Professor Mary Ryan.
Founded by influential tech leaders in 2015, OpenAI has become a driving force in the field of artificial intelligence research following the successful launch of the large language model ChatGPT in November 2022.
During the Keynote session, Zaremba recounted the story of how ChatGPT was created and reflected on questions around the democratisation of tech, and what large language models such as his own could mean for humanity.
"It's important that we engage with these critical developments in the field of artificial intelligence. The conversation with OpenAI is one that will help us to ensure that the work we do at Imperial spans both research and practice to ensure future fit ethical technologies." Professor Mary Ryan Vice-Provost for Research and Enterprise
Zaremba said: “We see a future in which AI plays a broader role in society” and highlighted that “every aspect of human life will be affected by AI”.
Both Zaremba and Dhar emphasized that OpenAI is hoping for a democratic process where, “as humans, we can decide how AI is used”.
Professor Mary Ryan said: "It's important that we engage with these critical developments in the field of artificial intelligence. The conversation with OpenAI is one that will help us to ensure that the work we do at Imperial spans both research and practice to ensure future fit ethical technologies."
Day One – Imperial addresses the technical foundations of large language models
Day One took place at Imperial, where DSI’s Co-Director Dr Mark Kennedy opened the symposium, and Lecturer of Computing at Imperial Dr Yingzhen Li, set the stage by giving a brief history of generative AI and large language models, focusing largely on attention mechanisms.
DLA Piper’s Jana Blount explained how generative AI and new large language models “seem like magic to the legal industry”, highlighting the opportunities that generative AI has to offer for law and regulation, as well as the challenges for this sector, namely issues surrounding data privacy and storage.
Following this, Imperial’s Professor Peter Pietzuch addressed the question of how we scale large language models and pointed out the “inherent unsustainability” of hardware advances, system design and environmental impacts currently present in this area.
Professor Aldo Faisal, Professor of Computing and Bioengineering, gave his talk on ‘Generative AI, from Data Analysis to Data Agency’, in which he described how “for the first time since fire, we are not the solution developer.” Following this, all the speakers engaged in a panel discussion around questions of power and responsibility.
Day Two – LSE explores the effects of LLMs on education, economics, finance, law and regulation
Day Two was opened by DSI Director at LSE, Professor Kenneth Benoit. Professor Benoit said: "AI is transforming education through changes in methods of learning and ways of teaching. It is vital that we address the policy issues affecting the university sector and that we embrace and prepare students for this future."
"AI is transforming education through changes in methods of learning and ways of teaching. It is vital that we address the policy issues affecting the university sector and that we embrace and prepare students for this future." - Professor Kenneth Bennoit, Director of the Data Science Institute at LSE
During the first session, Dr Jon Cardoso-Silva, Assistant Professorial Lecturer at the LSE DSI chaired a panel discussion on ChatGPT and the future of teaching, assessment and learning. This raised questions on how we can incorporate tools such as ChatGPT into the educational process and highlighted fears of malpractices, from students not consuming the output from ChatGPT, to academics not being able to distinguish between human and AI when marking assessments.
To encompass a range of views from across education, the panel was compromised of Professor Bingchun Meng from the Department of Media and Communications, Mark Baltovic, Senior Academic Developer at the Eden Centre for Education Enhancement, PhD student and member of the Working Group on Artificial Intelligence, Assessment and Academic Integrity Alexander Niederklapfer, and 3rd year undergraduate student in International Social and Public Policy at LSE, Idil Balci.
Professor Lilian Edwards from the Newcastle Law School addressed the challenges of generative AI for law and regulation, explaining how generative AI doesn’t fit into previous legal ideas of artificial intelligence.
In the final session of the symposium, Imperial’s Dr Mark Kennedy discussed the consequences of AI for the knowledge economy with Regius Professor of Economics and recipient of Nobel Memorial Prize in Economic Sciences, Professor Sir Christopher Pissarides, and Director at the Competition and Markets Authority, Hayley Fletcher.
Following the symposium, LSE hosted a separate event in conversation with Founder of the World Wide Web Sir Tim Berners-Lee, where LSE DSI Director Professor Kenneth Benoit and LSE President, Baroness Shafik discussed ‘How Web 3.0 Technology and Data Policy Combine to Promote Data Sovereignty, Privacy, and Trust.’
‘DSI Squared’ – Bringing together universities across STEM and the social sciences
The ‘Generative AI and The Knowledge Economy’ Symposium formed part of the ongoing collaborative initiative ‘DSI Squared’, which joins the Data Science Institutes from both Imperial College London and The London School of Economics and Political Science.
When it comes to data science research and its impact, the LSE’s strengths in the social sciences naturally complement Imperial’s strengths in science, technology, and medicine.
"It's been great to work with LSE on DSI Squared and we look forward to continuing this partnership." Dr Mark Kennedy Co-Director of Imperial's Data Science Institute
Imperial DSI Co-Director Dr Mark Kennedy said: "It's been great to work with LSE on DSI Squared and we look forward to continuing this partnership."
Apart from an ongoing seminar, networking series and small grant scheme, the symposium, which the team hopes to repeat annually, represents the largest achievement to emerge from the initiative so far.
If you are interested to find out more about the events, opportunities and activities happening as part of Imperial’s Data Science Institute, you can visit the DSI website, or sign up to the DSI Mailing List.
To find out more about LSE’s Data Science Institute, visit their website.
Photography by Stephen Swain and Robert Lawler.
Article text (excluding photos or graphics) © Imperial College London.
Photos and graphics subject to third party copyright used with permission or © Imperial College London.