Since last week, the NFB has been presenting a virtual reality (VR) installation designed by Sandra Rodriguez, which allows visitors to enter into dialogue with an artificial entity inspired by the famous American linguist Noam Chomsky.
Why Chomsky? This is the first question we ask Sandra Rodriguez, a trained filmmaker, who specializes in digital media research and who has taught for six years at the renowned Massachusetts Institute of Technology (MIT) – just like Chomsky until until recently, by the way…
“For several reasons,” she told us. First, because Chomsky is the most digitalized intellectual today. All the interviews he gives are royalty free. With all his writings, it represents a fantastic database for creating a conversational bot. »
The other reason that pushed Sandra Rodriguez to be interested in Chomsky is that the American linguist spent his life studying human intelligence through speech.
[Noam Chomsky] was a strong advocate of human creativity, which he believed could not be replicated [contrairement à ce que croient certains théoriciens de l’intelligence artificielle]but also the innate human capacity to ask questions, to be curious, and to collaborate with others.
Sandra Rodriguez, designer of Chom5ky vs Chomsky
This profile undoubtedly made him the ideal interlocutor to experience a moment of virtual dialogue. However, it is not necessary to know Noam Chomsky (or his work) to appreciate the experience offered by Sandra Rodriguez.
Let’s discuss
” What’s your name ? », Chomsky begins by telling us, before inviting us to ask him questions. You can ask him what he likes to eat for breakfast, but why not dive in like we did? Does he not believe that artificial intelligence (AI) can widen the gap between rich and poor? What does he think of transhumanism? Doesn’t AI inevitably lead to cheating? Chomsky’s answers are interesting and nuanced.
Every day, questions from the audience are dumped from the chatbot’s memory, but designers can see what topics were covered. It is whispered in our ears that last Wednesday, several visitors questioned Chomsky about the war in Ukraine. Not stupid…
The robot first draws from the database of digitized writings of Noam Chomsky. If there is nothing that could serve as an answer to your question in this bank, the system searches in an archive library of interviews and various other writings of Noam Chomsky generated by Chat GPT2. Finally, if there is still nothing for the robot to respond to you, the system draws on Chat GPT3.
Less inspired visitors, who perhaps have more difficulty formulating questions, will not be left out. Chomsky knows how to take the lead. He will ask you questions himself – mainly on artificial intelligence. Namely: what do we reproduce when we feed a conversational robot? What are we falsifying? What makes us unique? Good questions…
And the real Noam Chomsky?
By the way, is Noam Chomsky aware of this project? Will he come to Montreal to experience it?
Sandra Rodriguez actually made contact with the American intellectual in 2019. He was very interested in the project – the first incarnation of which was presented at the Sundance festival in 2020 – but health problems put an end to their trades. Subsequently, Chomsky’s entourage approved the stages of the experiment, but no, “at 94 years old, he will not put on a VR headset,” Sandra Rodriguez tells us.
Conversation in French is a bit bumpy, with the chatbot not catching all the words, and the simultaneous translation is not always happy.
It is certain that the AI systems are all trained in English, the data is mainly in English. So in French and German [parce qu’il s’agit d’une coproduction Canada-Allemagne]the system has difficulty recognizing certain words.
Sandra Rodriguez, designer of Chom5ky vs Chomsky
Sandra Rodriguez also highlights the irony of having a conversational robot played by Noam Chomsky, “who despite himself inspired the way chatbots are made [quand la conversion est fluide, on parle de grammaire chomskienne], but which has a fantastic legacy for human beings: we still know so little about the human brain, isn’t that wonderful? And isn’t it incredible to what extent humans are capable of being curious, collaborating and creating? »
This irony was brought home to her when she realized that the linguistics department at MIT was located just above the department where artificial intelligence was taught. “The students are in the same building, go to the same cafeteria, share the same vocabulary…”
Ultimately, Chom5ky vs Chomsky also has a social objective, Sandra Rodriguez tells us. “The promises of AI are not very healthy. Basically, we are told: it is inevitable, it is very complex and it will change everything. It’s not very reassuring… But the public has a choice. If we want to advance artificial intelligence in our lives, we must demystify the way it works. »
Now that Chom5ky vs Chomsky is launched, the production team hopes to tour the installation in museums and library spaces. Question of continuing this work of demystification.
Until October 15 at Espace ONF