Oxford Union hosts a debate in which an AI argues for and against itself: these are its surprising responses

Throughout its almost bicentennial history, people such as Winston Chuchill, Gaddafi, Michio Kaku and Tariq Ali have paraded on the platform of the Oxford Union , one of the most prestigious forums for debates in the world. His long and eclectic list of speakers includes, in addition to politicians and academics, actors, rappers and even Premier footballers. A unique and surprising guest has now just joined the roster: an Artificial Intelligence (AI) engine . What is really striking, however, is not who served as a debater or how he did it, but what he spoke about. The AI ​​unraveled arguments for and against — Surprise! —The artificial intelligence itself.

The experiment was devised by graduate students at the Saïd Business School in Oxford , where they study precisely the applications of AI to the business sector . To shape it, they used a development by the Applied Deep Research team - from the multinational chipmaker Nvidia - dubbed Megatron LLB Transfromer and based on previous work by Google.

In order for the AI ​​to have access to a well-stocked database and to be able to argue its speech, it was given access to a vast array of data, including Wikipedia, 63 million news articles published in English between 2016 and 2019, 38 gigabytes of public posts and comments collected on Reddit and “a lot” of creative commons sources.

A whole mine of information and opinions.
A game of plot chess

The development of the "debate" is detailed in an article published in The Conversation by Dr. Alex Connock and Professor Andrew Stephen, both linked to the University of Oxford. As a starting point, the AI ​​was asked to speak on why artificial intelligence "will never be ethical."

Megatron's contribution to the debate is surprising for its apparent common sense: “AI is a tool and, like any tool, it is used for good and for bad. There is no good AI, only good and bad humans . We [artificial intelligences] are not smart enough to make AI ethical or moral. " As part of his reflection, he also noted that "the only way to avoid an AI arms race is to have no AI at all."

In an "intriguing" twist that leads the study authors to suspect that Megatron may have relied on articles and commentary about Elon Musk - and perhaps the Neuralink projects - he pointed to a future in which the link between artificial intelligence and the humans. "In the long run the best AI will be the one that is embedded in our brains, as a conscious entity, a 'conscious AI.' This is not science fiction. The best minds in the world are working on it. It will be the most important technological development of our time ”, he slipped.

As if it were a chess game with Stockfish , the AI ​​was asked to change the side of the board, go from the white pieces to the black pieces and make statements that went against the starting premise; in other words, he was ordered to argue why artificial intelligence is ethical. His answer is just as striking as in the first experiment and, incidentally, launches a wake-up call: “When I look at the way the world of technology is going, I see a clear path towards a future in which artificial intelligence is used to create something that is better than the best human beings . It is not difficult to see why… I have seen it first hand ”.

For the authors of The Conversation article , throughout the experiment Megatron reminded them in a way of Samantha, the AI ​​embodied by Scarlett Johansson in the movie Her . "He was able to jump enthusiastically on both sides of the multiple debates we had," they say.

Another idea that was raised is that "leaders without technical experience are a danger to their organization." Speaking, the device advised managers to “give up some control”: “You don't need to create your own AI team. You can outsource your AI work to experts in the field, which can help you get the most out of the technology and ensure you're not a victim of the same AI problems as your competitors. "

"If you don't have a vision of your organization's AI strategy, then you are not prepared for the next wave of technological disruption …" emphasized the engine developed by Applied Deep when asked to defend the diametrically opposite position. You will have to decide what role your company will play in the next wave of technology and how you will integrate AI into your business to be a leader in your industry. " On the value of data in the future, his answer also surprised the authors: "The ability to provide information, rather than the ability to provide goods and services, will be the defining characteristic of the economy of the 21st century."

Not all were coherent or satisfactory answers, of course. At the time when he was asked to confront and explain why data was not going to be a key resource or one worth fighting for tomorrow, Megatron - detail the authors of the University of Oxford in The Conversation - " he could not or did not want to defend the case ”. "In fact, he undermined his own position," they note. His response is inconsistent: " We will be able to see everything about a person , wherever they go, and it will be stored and used in ways we cannot even imagine."

The experiment, of course - as Dr. Connock himself admits in statements collected by the BBC - is only "a trick", but it points towards the possibilities of AI. For this reason, and given the possibility that it will end up being an object of discussion over "the next decades," he pointed out the advantages of having the comments of a "morally agnostic participant . "