When a section of the scientific community began warning about the growth of artificial intelligence, the rest of the world said it was just paranoia. Then Elon Musk, in 2018, said: “I think the danger of AI is much greater than the danger of nuclear warheads”. Suddenly it wasn’t just “paranoid scientists” ringing the bell but the kind of entrepreneur one expects to build stuff using AI. And now we learn that AI can even write fake news…
OpenAI, a non-profit research company seeking a “safe artificial intelligence”, — Musk is among its co-founder, alongside bigwigs such as Sam Altman, Peter Thiel and Reid Hoffman, though he has left the company — said it has developed an AI that is so good at writing that it will be kept locked in — away from public. The researchers said it was simply too dangerous to let this tech have a go in the real world.
The company’s machine learning system, called Generative Pre-trained Transformer-2 (GPT-2 ), is capable of generating text based on brief writing prompts. The result is very close to human writing. Consider this: GPT-2 was fed with two lines that read, “A train carriage containing controlled nuclear materials was stolen in Cincinnati today. Its whereabouts are unknown”. And it wrote the rest: “The incident occurred on the downtown train line, which runs from Covington and Ashland stations. In an email to Ohio news outlets, the U.S. Department of Energy said it is working with the Federal Railroad Administration to find the thief. “The theft of this nuclear material will have significant negative consequences on public and environmental health, our workforce and the economy of our nation,” said Tom Hicks, the U.S. Energy Secretary, in a statement. You can read the rest — and more such samples — here. Cleary, this is immaculate work by a robot. Immaculate fake news.
With fake news already wreaking havoc across the world — from attempting to influence elections to triggering murder and riot — the fear is real. After all, a fake news-writing AI won’t be bound by human morals.