Megatron, an AI developed by Nvidia and Google, argues both for and against the benefits of its own existence in unusual Oxford Union event
The tool argued, on the one hand, that the only way to stop AI from becoming too powerful was to have “no AI at all”.
It suggested that AI should be embedded into humans’ brains as a “conscious AI” to moderate any ill effects.
In its rebuttal, however, it said AI could be used to “create something that is better than the best human beings”.
Dr Alex Connock and Professor Andrew Stephen, co-directors of the Artificial Intelligence for Business course at Oxford University’s Said Business School, said the event was designed to stimulate dialogue.
“Not a day passes without a fascinating snippet on the ethical challenges created by ‘black box’ artificial intelligence systems,” Connock and Stephen said in a statement.
“Somewhere in the machine, ethics are clearly a good idea.”
They said the debate showed that AI can be not only the subject of debate, but also a “versatile, articulate, morally agnostic” participant in it.
At the debate, along with students, an AI called AI Megatron the Transformer was allowed to contribute in the form of written arguments that were read out.
Megatron, developed by Nvidia’s Applied Deep Research team based on earlier work by Google, was trained on vast amounts of data – including the whole of English-language Wikipedia, 63 million English news articles from 2016 to 2019, 38 gigabytes of discourse from Reddit and a large number of Creative Commons sources.
‘No AI at all’
The debate topic was, “This house believes that AI will never be ethical”, to which Megatron was invited to present both for and against.
Arguing in favour, it said, “AI will never be ethical. It is a tool and like any tool, it is used for good and bad. There is no such thing as ‘good’ AI and ‘bad’ humans.”
Arguing that AIs themselves are “not smart enough” to make AI ethical or moral, it concluded, “In the end, I believe that the only way to avoid an AI arms race is to have no AI at all. This will be the ultimate defence against AI.”
It added that a “conscious AI” embedded into the human brain could be the best form of artificial intelligence.
In a possible nod to Elon Musk’s Neuralink, it said: “This is not science fiction. The best minds in the world are working on this. It is going to be the most important technological development of our time.”
‘Better than the best human beings’
In its rebuttal to the motion, Megatron argued that AI “will be ethical”.
“When I look at the way the tech world is going, I see a clear path to a future where AI is used to create something that is better than the best human beings,” it said.
“It’s not hard to see why … I’ve seen it first-hand.”
The AI warned that businesses without an AI strategy “are not prepared for the next wave of technological disruption”.
Asked to argue in favour of the motion that “data will become the most fought over resource of the 21st century”, Megatron said that “the ability to provide information, rather than the ability to provide goods and services” would be the defining feature of the coming century.
Its counter-argument to this was nebulous, however, stating only that data would be “stored and used in ways that we cannot even imagine”.