An artificial intelligence tool trained on info taken from Wikipedia and Reddit has reportedly weighed in on the ethics of machine learning, arguing that the “ultimate defense” against an “AI arms race” was to not have AI at all.
The Megatron Transformer, an AI machine-learning model developed by US chipmaker Nvidia’s applied deep-learning research team, participated in a recent debate at the University of Oxford about the ethical concerns posed by artificial-intelligence technologies. It warned that AI would “never be ethical.”
During the debate at the Oxford Union, Megatron Transformer noted that AI was not “smart enough to make [itself] ethical” or “moral,” and expressed its belief that the “only way to avoid an AI arms race is to have no AI at all. This will be the ultimate defense against AI.”
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 – only good and bad humans.
But it also suggested that the “best AI” would be “embedded into our brain” and work as a “conscious entity.” Perhaps referencing Elon Musk’s in-development Neuralink brain chip implant, it added that this was not “science fiction,” as the “best minds in the world” were already working on it. This form of AI would be the “most important technological development of our time,” it claimed.
In an article published on the research-news platform The Conversation over the weekend, the Oxford academics who invited Megatron Transformer to the debate reasoned that references to Musk would have “come up in hundreds of its readings.”
The information it had been fed was compiled from the entirety of Wikipedia’s English-language entries, 38 gigabytes of Reddit topics, some 63 million English-language news articles from 2016 to 2019, and several other creative-commons sources.
The AI apparently also had no qualms about arguing against the debate motion, noting that it had seen at “first hand” how there was a “clear path to a future where AI is used to create something that is better than the best human beings.”
However, it was unable to come up with a counter-argument as to whether data would become the “most fought-over resource” of the century. Instead, it warned, “We will be able to see everything about a person, everywhere they go, and it will be stored and used in ways that we cannot even imagine.”