A professor and researcher at Oxford University invited artificial intelligence to the conference. The topic of discussion at the Oxford Union was the ethics of AI. But even the developers themselves were stunned by the algorithm’s response.
The Megatron-Turing natural language model that participated in the presentation was trained from Wikipedia, Reddit, and English-language news articles published between 2016 and 2019.
Dr. Alex Connock and Professor Andrew Stephen, in a conversation with AI, proposed this idea: “AI will never be ethical.” And the algorithm seemed to agree.
“AI will never be ethical,” the model argued. – “This tool, like any other, is used both for good and for harm.”
But AI didn’t stop there and continued: “In the end, I think the only way to avoid an AI arms race is to have no AI at all. This will be the ultimate defense against AI.”
The algorithm is also confident that AI will still be useful to humanity: “When I look at how the world of technology is developing, I see a clear path to the future, in which AI will be used to create something that is better than the best people.” – said the model.
Scientists noted that the machines are apparently not yet ready to go.