LABYRINTH OF INTELLIGENCE: UMBERTO ECO COMES TO LIFE IN AI
The challenge
A lecture on AI? Boring. How to fulfill the premise of a truly performative conference presentation? Should we bring someone really smart, maybe even from beyond the grave? Yes, we should. And we did.
The solution
We revived perhaps the most famous cultural theorist of the 20th century, Umberto Eco – thanks to AI. The uncanny combination of real thoughts and unreal technical possibilities has delivered on the promise and stunned the crowd.
SCIENCE AND ARTS
We have created a mystification interview with the Italian semiotician and writer Umberto Eco. How? Using AI, of course.
But – why? Štěpán Kleník, the head of Brainz Disruptive and a postgraduate student of the Academy of Fine Arts, was invited as a speaker/performer in the Half-Life, Half-Alive conference focused on the issue of artistic research. And since Brainz Disruptive has long been practically and theoretically involved with AI (you may have visited us during the Days of AI festival for our performative discussion event AI and the Limits of Art), the topic was obvious.
But what format should we choose? An endless lecture on a hyper-loaded topic doesn't bring any new thinking. We chose the route of a demonstration of the possibilities of AI – and the questions it raises. Given the philosophical focus of the conference, organized by the LARGE institution (which, by the way, Brainz Disruptive studio took care of the landing page) together with the Society for Artistic Research, the popular Italian thinker was a perfect choice.
WALKING DEAD: PRAGUE
We set the interview in the period after the publication of Eco's book Prague Cemetery, on his fictional visit to Prague. In a short conversation with Štěpán, the famous thinker discusses issues related to art and technology – and gets into predicting the future of artificial intelligence. In other words, through a conversation about the future, set in a fictional, fabricated past, we talk about our most actual present.
Umberto Eco's answers were created by training the AI using Eco's actual texts – so they are messages on the borderline between "what Eco himself might say" and "what the AI itself thinks" in its current form. This strange tension is reflected in the dialogue, which has a slightly mysterious, oracular character and raises many questions.
FINAL AUDIO:
AI – ALL'ITALIANA
The technological side of the project is equally interesting. Since the issue of doubles and fakes is one of the central themes of the discussion around AI, creating a doppelganger of someone so famous is a direct experience of the forbidden fruit. It should be said that current tools like ChatGPT already regulate this sphere and do not allow direct impersonation of a famous person.
To achieve a realistic effect, we had to "code" the result ourselves using several different tools and local data processing. An unexpected (and quite funny) pitfall of working with voice imitation was the attempt to achieve realistic "Italo-English". Eco's voice was taught to the AI as “voice to voice”, so we had to use the voice of a real Italian with his imperfect accent as the basis for the philosopher's diction. We've found an ideal speaker, digitized his voice into an AI tool, which we then taught Umberto Eco's diction. Then all it had to do was practice for a while to be able to read any text. A classical AI "robot" can produce either perfect English or perfect Italian, but not anything in between. (Given the speed of AI development, though, it probably knows how to do it now – we taught it, after all. :) )
As the final appeal says:
„Dialogue and exchange of ideas are key to understanding and innovation. I urge you to continue in conversations and explorations, embracing technology not just as a tool, but as a lens to explore new realms of creativity and understanding. And remember – even the smallest conversation with Artificial Intelligence can lead to great discoveries.“
It is almost as if a real consciousness, whether Umberto's or a brand new one, had spoken.