This week, an Italian artist and programmer named Giacomo Miceli launched The Infinite Conversation website, an AI-powered non-stop chat between artificial versions of German director Werner Herzog and Slovenian philosopher Slavoj Žižek, complete with realistic voices.
Visiting the site, which is not affiliated with either person, you will see AI-generated charcoal portraits of the two men in profile. Between them, a transcript of the AI-generated text is highlighted in yellow as AI-generated voices simulating those of Herzog or Žižek read it. The conversation goes back and forth between them, with distinct accents, and you can jump between each segment by clicking the arrows under the portraits.
Its creator is positioning the site as a social commentary on audio deepfakes and upcoming technologies that could undermine trust in the media in the near future. “This project aims to raise awareness about the ease of use of real voice synthesis tools,” Miceli writes on the site. “Right now, any motivated fool can do it with a laptop in their bedroom.”
Herzog and Žižek appear to be particularly ripe targets for AI impersonation, as listeners might be predisposed to believe that the philosophical, philosophical director might be saying deep, hard-to-understand things. As a result, when the grand GPT-3 style language model behind The Infinite Conversation spews philosophical nonsense, it almost feels like the real thing. Here is an example of something the fake Herzog said on the site:
In a way, Freud also has something to do with literature.
He was, after all, a writer.
Yes, he was a scientist, he wanted
being a scientist, but he was also a writer
who wrote these strange stories.
There is something that seems contradictory in Freud.
On the one hand, he had such
strong anthropological vision, which I find very appealing
on the other hand, he was limited in his understanding of cultural history.
He was very antiquarian, for him antiquity was
the most important era because it clearly revealed the impulses
whereas the Middle Ages were just mean.
It was only towards the end of his life that
he sees nothing good in the Middle Ages.
The dialogue apparently drags on. “When you open this website, you are taken to a random point in the dialogue,” Miceli writes. “Every day a new segment of the conversation is added. New segments can be generated at a faster rate than it takes to listen to them. In theory, this conversation could continue until the end of time.”
Miceli is said to have created the site using “open source tools available to everyone”, declining to give technical details, although he wrote on Hacker News that he could create an explanatory article within the next week. “The generation of the script itself is done using a popular language model that has been honed on interviews and content written by each of the two speakers,” he wrote in the site’s FAQ.
On Ars, we’ve already covered technology that can manipulate your voice using AI or even allow someone to impersonate someone else audibly. And in October, we saw a podcast that featured similar text-to-speech technology to power a fake interview between Steve Jobs and Joe Rogan.
Few doubt that we are at the forefront of a new era in synthetic media, but when the machines speak, will they make sense? On The Infinite Conversation, that’s not quite the case yet. “Everything you hear is entirely machine generated,” writes Miceli. “The opinions and beliefs expressed do not represent anyone. These are the hallucinations of a silicon slab.”