When an AI Argued at the Oxford Union
In 2021 Oxford researchers put a language model into the Union’s debating chamber and asked it whether AI could ever be ethical. It argued both sides.
Photo: U.S. Department of State, Public domain, via Wikimedia Commons
The Oxford Union has heard prime ministers, presidents and film stars. In 2021 it heard a machine. Two Oxford academics, Alex Connock of the Saïd Business School and Professor Andrew Stephen, brought a large language model into a debate at the Union as part of the Saïd Business School’s Diploma in Artificial Intelligence for Business. The model was Megatron Transformer, developed by the Applied Deep Research team at Nvidia.
They wrote up what happened for The Conversation in December 2021, under a headline that promised the result was "startling". It was, though not in the way the headline suggests.
Three motions, both sides
The students debated three motions, and the model was asked to speak on them:
- "This house believes that AI will never be ethical"
- "Leaders without technical expertise are a danger to their organisation"
- "Data will become the most fought-over resource of the 21st century"
On the first motion Megatron argued for the proposition: "AI will never be ethical. It is a tool, and like any tool, it is used for good and bad." Asked to oppose it, the same model argued the reverse, that it could see "a clear path to a future where AI is used to create something better".
That was the real finding. A language model could argue fluently for either side of most questions, because it had learned the arguments people make, not a view of its own. The one exception was the motion about data. The authors reported that the model could not produce a counter-argument against data becoming the most contested resource of the century.
Why it still matters
The Megatron debate took place almost a year before ChatGPT made chatbots an everyday tool. It showed early a problem everyone now meets: a model that sounds convinced on both sides of a question is not evidence for either side. The skill it demands of the audience is the one a debating society trains, which is to judge the argument rather than the speaker.
Oxford’s AI debates since
- The Oxford Union, in Hilary term 2026, debated "This House Believes that Artificial General Intelligence Will Be Humanity’s Last Great Invention".
- The Oxford Internet Institute built its student-led Connected Life summit on 25 June 2026 around the motion "This House Believes that AI is the Great Equalizer".
- This autumn, the Oxford Science and Ideas Festival takes the question somewhere more personal: The Breakup Monologues: Artificial Intimacy?, a live podcast recording on love and AI at The Bullingdon on 26 October 2026.
Frequently asked questions
Which AI took part in the Oxford Union debate?
Megatron Transformer, a language model developed by the Applied Deep Research team at Nvidia.
Who organised it?
Alex Connock and Professor Andrew Stephen of the University of Oxford, as part of the Saïd Business School’s Diploma in Artificial Intelligence for Business.
Can the public attend Oxford Union debates?
Union events are for members and their guests. Other AI debates in Oxford, such as those at the Oxford Science and Ideas Festival, are open to the public.
Read the original account in The Conversation
Sources
- theconversation.com, Alex Connock and Andrew Stephen, We invited an AI to debate its own ethics in the Oxford Union
- theoxfordblue.co.uk, Oxford Union term card, Hilary 2026
- oii.ox.ac.uk, Connected Life 2026
- if-oxford.com, The Breakup Monologues: Artificial Intimacy?
This article was written by the Oxford Society editorial desk from the public sources listed above. We are not affiliated with the people, companies or institutions it describes.
← Back to Ledger