About this event:

The first event in the Technomoral Conversations series - join us to explore the Ethics of Large Language Models.

When Artificial Intelligence Speaks: A Conversation on the Ethics of Large Language Models

Admission: Free, open to all, online event.

In a recent paper that received widespread media coverage, a team of computing researchers compared the latest AI trend of large language models (LLMs) to parrots, because of how these powerful new tools (such as OpenAI’s GPT-3 and Google’s BERT and Switch-C) mimic and remix human language use.

Some media outlets have championed these technologies as the next wave in AI that may outcompete humans, by generating endless new stories, essays, news items, poems and even songs on command. Critics, however, have pointed to ethical risks with these new tools, for example, evidence that current LLMs are even more prone to generating falsehoods than smaller models, and far more likely to do so than humans.

So what does this new trend in AI mean for us? Is there hope that these silicon ’parrots’ can be tamed, that LLMs can be made safe and reliable new companions to human knowledge and culture?

On Tuesday December 7th at 5pm, join us as we explore these questions in a free-ranging conversation between our Centre for Technomoral Futures Chancellor’s Fellow, philosopher Dr. Atoosa Kasirzadeh, and Iason Gabriel, Staff Research Scientist at DeepMind, followed by questions from the student and public audience. The event is the first in a series of Technomoral Conversations hosted at the Centre, which aim to bridge the increasingly harmful divide in our society between technical and moral knowledge.

Panellists:

Dr. Atoosa Kasirzadeh joins the University of Edinburgh’s Centre for Technomoral Futures and the Department of Philosophy as a Chancellors Fellow in November 2021, following a visiting research post at DeepMind in their ethics and society team. She holds a PhD in philosophy of science and technology from the University of Toronto and a PhD in applied mathematics from École Polytechnique de Montréal. Her research focuses on the ethics and philosophy of AI and computing, the implications of mathematical and computational modeling of the socio-economic world, and values in sciences and decision making.

Iason Gabriel is a staff research scientist on the ethics research team at DeepMind. His work focuses on the intersection between Artificial Intelligence and ethics, with a particular focus on the relationship between technology and human values. Before moving to DeepMind, Iason taught ethics and political theory at Oxford University. He also worked for the United Nations in Sudan and Lebanon.

About the Series

As part of the Centre’s Baillie Gifford Research Programme in the Ethics of Data and Artificial Intelligence, the Technomoral Conversations series will host bimonthly dialogues on today’s hard questions about living well with technology. Moderated by Centre Director Shannon Vallor, these conversations bring together the diverse forms of technical and moral expertise held by academic researchers, industry professionals, community representatives and advocates, artists, students and voices from the public and third sector.

Have Your Say: Want to suggest a topic for our series? Do you want to propose a contributor you’d like to hear from in our series? Or just want to join our mailing list to hear about future events in the series? Email us at ctf@ed.ac.uk

Have questions about this event?

Please email us at ctf@ed.ac.uk

Joining link and the platform

The joining Zoom link and instructions will be shared 24hrs ahead of the event via email.

Q&A session

During the event you will be able to ask questions using the Q&A function. There will be no microphones or web cams available for viewers at this event. To ask a question via this method use the Q&A function at the bottom of your screen. Type your question in the box and press send. If you want to ask your question anonymously please select the box, ask anonymously.

Please note: The Q&A will be moderated and we may not have time to answer all questions.

Accessibility

The event will be supported by British Sign Language Interpreters. If you need any other arrangements to be made for you to be able to attend the event, please let us know at the time of the registration or via email at ctf@ed.ac.uk.