The creators of a revolutionary AI system that can write news stories and fiction works – called "deepfakes for text" – have taken the unusual step of not releasing their research publicly for fear of potential abuse.
OpenAI, a nonprofit research company supported by Elon Musk, Reid Hoffman, Sam Altman and others, says that the new AI model, called GPT2, is so good and that the risk of malicious use is so high that it breaks off its normal practice of releasing the full research to the public to give more time to discuss the consequences of the technological breakthrough.
When used to generate new text, GPT2 is able to write plausible passages that match what is given in both style and subject. It rarely shows any of the quirks that mark earlier AI systems, such as forgetting what it writes about midway through a paragraph, or missing the syntax of long sentences.
Feed the opening line of George Orwell's Nineteen Eighties ̵
"I I was in my car on my way to a new job in Seattle, I put in the gas, put the key in, and then I ran it. I thought exactly what the day would be. A hundred years from now. or another school in a bad part of rural China. I started with Chinese history and science history. "
Give it the first few paragraphs of a Guardian story of Brexit, and its production is plausible newspaper process, filled with" quotes "from Jeremy Corbyn, mentions the Irish border, and answers from the prime minister's spokesman.
One such, quite artificial, paragraph reads: "Requested to clarify the reports, said a May spokesman:" PM has made it clear that she intends to leave the EU as soon as possible and it will be under her negotiating mandate confirmed in the Queen's speech last week. & # 39; "