Former world chess champion Garry Kasparov knows firsthand what it feels like to compete against artificial intelligence.
In 1997, Kasparov played two rounds of chess against IBM’s Deep Blue supercomputer–winning the first before famously losing the second—and has since then gone on to write a book about the future of AI and what it means for humanity. (The book, Deep Thinking: Where Machine Intelligence Ends and Human Creativity Begins, 1st Edition, was released in 2017.)
During South by Southwest 2019 in Austin, Texas, Kasparov was in town for a discussion series hosted by Sony about AI and creativity. Afterward, he sat down with Adweek to talk about those topics, along with the future of data privacy and what it means for artists, governments and everyone in between.
Photo: Dianna McDougall
“I called it in 2015 that the Facebook business model—after learning what I knew about Putin and the troll factories and the fake news industry—I said the Facebook business model was like a beehive for a Russian bear,” he said. “Because it was tempting to just eat. And I think we should learn more about it. We’re dealing with a real existential threat, while dictators and quasi-state organizations are going after us and using technology created in the free world to undermine the very foundation of the free world.”
This interview has been edited for brevity and clarity.
Adweek: Do you think AI can ever be truly artistic?
Kasparov: It’s a slippery ground of semantics. Because the answer to your question is of course yes, because artistic means what? Will you find enough people that could enjoy it? Absolutely. I think a majority will not, but it’s like a movie production. What makes it really artistic? Spending a lot of money? I don’t think so. Casablanca was shot in a couple of rooms and it’s one of the best movies ever made.
And you have a few movies that spend hundreds of millions of dollars and they just didn’t last. So you always have people who might enjoy it, and if you have 1 million people who enjoy it and 1 billion people who didn’t and who think it’s garbage, does it make it artistic? I think yes. The moment we talk about something that doesn’t have a very clear scale relation or that’s widely acceptable, the answer is yes. So AI could come up with a movie and maybe someday this movie will win an Oscar because people think ‘Oh, it’s sensational. It’s fine. Let’s do it for the sake of diversity.’ An AI can’t operate in the area of human creativity because it’s so subjective. I think it will not be Shakespeare, but it will always have its audience.
With computers so reliant on creating based on data, can an AI ever be creative from scratch on its own?
I don’t think so, because machines should know the odds. You know, it brings us back to another philosophical issue that was raised by Joseph Weizenbaum, one of the founding fathers of AI, in his 1976 book Computer Power and Human Reason. When he talked about the subtle difference between choosing and deciding, he argued that deciding was computational, because always you know you go deep, deep, deep, down to the bottom and find “I was told,” when choosing is “I want it.” And I think creativity, human creativity, is “I want it.” There’s no reason for that; it’s just “I did it.” I always like quoting Picasso, who said machines are useless because they only give you answers.
An AI can’t operate in the area of human creativity because it’s so subjective. I think it will not be Shakespeare, but it will always have its audience.
Creativity starts with a question; questions are the beginning. So if you want to invent or reinvent yourself as an artist, you have to ask questions. And machines can ask questions, but I’m not sure they know which questions are relevant. So, again, people say it’s ‘you say, they say,’ but it seems human creativity is where knowing the odds is counterproductive. Creativity doesn’t mean that you’ll be successful. You can be creative and fail, and if you knew the odds, you’d never do that. Creativity means you don’t know the outcome. It’s something new, something disruptive. It may fail. And a machine will never do that because it’s just the odds will never be in its favor.