Anthony Horowitz, the prolific novelist and screenwriter, has confessed to being a frequent ChatGPT user despite his fears that it may be cheating.
Horowitz said the technology was useful for research that he would previously have done in a library. However, he added that its findings could not be entirely trusted and admitted it felt like cheating.
“I feel nervous about it because, is it cheating? I feel a bit like it’s cheating in a school exam,” he told The Independent.
The advance of artificial intelligence has created a series of dilemmas for the cultural world. There has been widespread anger at the “theft” of copyrighted material by organisations, including Meta, to train AI systems, leading to a number of court cases alleging copyright infringement. Last year the AI company Anthropic paid £1.1 billion to settle a class action lawsuit filed by writers who claimed the firm had stolen their work to train its large language models.
Publishers have said that an increasing number of writers are using AI tools to write manuscripts. They have warned that as the sophistication of AI tools increases, it will become harder to identify whether works were created by humans.
Horowitz said in the interview this week that the limitations of the technology needed to be understood and gave the example of when he had asked ChatGPT for an adjective to describe the shape of a potato. It returned the word “ellipsoid”.
“This is when you drop the artificial intelligence and use your own intelligence. Readers hearing that the potato on her plate was ‘ellipsoid’ are going to say, ‘What the hell is this guy talking about?’” the 71-year-old writer said.
“It’s useful to have immediate knowledge but not unrestrained. You’ve always got to judge what you’re being told and also look for a second source,” he said, adding: “In the old days I’d go to a library. I’d look up stuff in books. What’s the difference between using AI and going straight to the point?”
In an interview with Times Radio on Wednesday, Horowitz said that while the tools could be good for research, they were “dangerous friends”.
He said that he had not used AI to write any of his novels, adding: “People can tell books that have been written by AI simply because AI isn’t very good. It can’t emote, AI can’t poke fun at itself, AI has no sense of irony.”
Horowitz was awarded a CBE in 2022 Victoria Jones/PA
He added that he thinks “even a sentence of AI would sit in my book as like something floating on a pond, as something you don’t really want to see when you’re having a swim”.
“It just doesn’t work. It’s a different world. And so it is, as I say, a friend, but a dangerous friend and one to be kept at a distance.”
Although Horowitz has admitted to using AI, he has turned his back on other elements of the “new world”.
“I am free of all social media and feel so much healthier and better for it. The change is really quite dramatic,” Horowitz said after revealing that he left X two years ago. Horowitz said that he had found himself “disappearing into it” and had been encountering “noticeably more toxic content”.
The author keeps up with the news — he wrote a letter to The Times this week criticising the comments of the former ambassador Sir Tony Brenton, who called upon the UK’s Jewish community to be more critical of the Israeli government — but he prefers starting the day in a more artistic way. He begins each day with a poem, saying that he finds it “endlessly encouraging and life-enhancing”. Horowitz did not say whether he was a fan of poetry written by AI.