These artists are exposing the dangers of AI and surveillance through art
Two artists use their creativity to reveal issues in modern technology
The ubiquity of surveillance and artificial intelligence is often considered evidence that society is becoming a dystopia. And two artists are using their creativity to illustrate the potential dangers of these technologies.
Giacomo Miceli is a computer scientist and the creator of "The Infinite Conversation," which features documentary filmmaker Werner Herzog and Marxist philosopher Slavoj Žižek in a never-ending discussion.
Miceli scours the internet for actual recorded interviews with the men, and has trained an AI to speak in their voice.
"Every day is a new piece added to this conversation. I generate about half an hour or one hour, depending on the day of conversation, and it's appended at the end. And this will continue happening until I stop paying the server bills or society falls in some order," he told Spark host Nora Young.
The point is to try to demonstrate the possible misuse of machine learning, and how it can be abused to make nearly anyone appear to say anything.
"It meant going through audiobooks, going through interviews, isolating the portion in which the person you're interested in recreating [is in], cloning the voice and isolating that portion and fitting it into the model. And then something similar is done also for the text," Miceli explained, describing the process used to build "The Infinite Conversation."
Although he appreciates the power of machine learning, and is optimistic about its potential, he also has deep concerns.
"The fact that it's so easy to generate those voices makes it a little bit unsettling. The notion that what we hear online, and what we consume online, can be manufactured in a way that is not intuitive, for us humans to understand, is very worrisome to me. I just think that this can be misused in so many ways. And I was happy to breach those rules in a playful way."
Flemish artist Dries Depoorter is doing a similar project, but rather than exposing the risks of machine learning, he hijacks public security cameras around the world to expose the risks of our Instagrammable moments.
He was inspired to take on the project by watching people taking selfies in public spaces. He designed a machine learning algorithm to find security camera footage of people preparing to take selfies and then designed another algorithm to find the actual selfies online. In this way, it was easy for him to demonstrate just how much personal information can be scraped from a simple picture posted online.
"I'm talking about security cameras, cameras that are sometimes even in private spaces. But they're all open or have a standard password. So what the standard password I'm seeing, it's '1234', or 'admin'. I'm not going to show a private camera in a hospital, but I try it in a creative way."
He is keen to point out that he is just one person working by himself with limited access to hacking tools – and that he's really trying to demonstrate the potential of data and identity theft with facial recognition.
"Imagine what the government can do," he said.
Depoorter hopes his work will make people better understand the risks they take even with the simple act of posting a selfie to Instagram.
"I try to show the dangers of technology with my work. And for me, it's important that it's playful and a lot of people can understand it," he said.
Miceli agrees: "It is very hard to predict exactly how this will be used and how we will react as a society. But I think the first and most important thing is just to be aware."
Written by Adam Killick. Produced by Nora Young, Samraweet Yohannes and Olsy Sorokina.