We caught up with director Piotr Mirowski to hear all about Improbiotics’ RoboTales at Gilded Balloon Patter House, an improvised tech theatre show fusing human creativity and artificial intelligence. For tickets go to https://www.edfringe.com/tickets/whats-on/improbotics-presents-robotales
Where did the idea for mixing improv theatre with AI come from?
We create shows that connect audiences with experimental technology, engage audiences about AI ethics, perform grounded theatrical improvisation and entertain festival punters looking for comedy!
I have two passions: theatre and robots! The idea of mixing these two fields came to me about 10 years ago, when I was researching and building AI models, while also performing on stage as an actor and improviser. I noticed a strange analogy between the large language models I was coding up (which, as we know, keep making stuff up and generate the most likely answer given their training data and context) and improvisers (who strive at being spontaneous, and at always saying “the most obvious” thing, while also carefully listening to their stage partners). Spontaneity is surprisingly hard for adult humans, because we’ve been socialised to think twice before we speak, and it was interesting to me that a mindless AI did not have that problem as it would always give an answer – no matter how wrong. So I decided I had to try to improvise on stage with a robot, powered by a language model that I designed from scratch. Obviously, the analogy between machines and improvisers ends there, and the idea turned out to be quite a challenge.
In 2016, I met Kory Mathewson, another researcher in robotics and an improv comedian at the Rapid Fire Theatre in Edmonton, Canada, who had the same idea. Kory and I became friends and set up to create a theatre company, called Improbotics, where human actors raise up to the challenge to improvise alongside machines. Our ethos is to make your stage partner look good, even when it is a robot! Over the years, the Improbotics shows evolved, thanks to the many new friends we made and to the large international cast of humans who came to shape the company. Boyd Branch designed augmented reality shows that kept us connected and provided emotional support during Covid lockdowns. We improvised multilingual shows with live AI translation with the Swedish cast of Improbotics. We explored AI as a tool for grounded, emotional improv with Sarah Davies.
What’s it like working with A.L.Ex on stage? Does the robot ever surprise you?
It does, although not always for the right reasons! At the beginning, in 2016, when AI was not as advanced as today, our robot A.L.Ex would simply generate non-sequitur responses that were sometimes hard to work with. I integrated more powerful AI models inside A.L.Ex’s code, that could better remember the context of a scene. Nevertheless, with other Improbotics cast members, we found it was like improvising “with an X factor”, or “with a very novice / nervous improviser who would shout random suggestions”, and the hard work was on the human side, trying to justify those offers.
We replaced the robot with a human “Cyborg” – an actor who takes their lines from AI via an earpiece or via augmented reality glasses displaying AI-generated lines. That added physicality, emotional expression, interpretation and subtext. Our system could be seen as an actor training technique of sorts, with actors cold reading seemingly absurdist text. When we were using earpieces, it was delightful for the actors to start speaking a sentence without knowing how it would end!
In 2023, when we started using one of those publicly available chatbots under the hood of A.L.Ex, the AI suggestions became more bland. It was a negative consequence of tech companies trying to make these chatbots “safe” by making them “honest, helpful and harmless” – but not necessarily useful in a creative context. So once again, it fell upon the human actors to do the leg work, and to provide A.L.Ex with interesting and very specific context. We are sometimes surprised when A.L.Ex says something spicy that was not in our dialogue.
This year, we are bringing a new show to Edinburgh: RoboTales. It is an improvised choose-your-own adventure game, with audience participation. The bot uses speech recognition and a piece of software I spent months developing, to analyse the improv scenes and generate new choices and scene transitions. The audiences control the story by voting on their phones for their preferred choice (like a modern take on the choose-your-own adventure books), and then the actors take the story forward. It is like “Black Mirror: Bandersnatch”, brought to life by AI and the actors’ imagination and improv talent.
How much of each show is planned—and how much is pure chaos?
We are telling a story about our relationship to technology and of the place we are giving it, and at the same time the show progressively builds up the tech. We start with direct interactions with the robot, then with Cyborg actors, then deep fakes and made-up science presentation, until we finish with the long-form choose-your-own adventure. As the show progresses, the technology retreats and we focus more on the actors’ performance. So we have a narrative structure, and the chaos comes from the collision of human and AI suggestions.
What’s been the funniest or strangest moment the AI has created so far?
All the instances where AI created something strange and interesting, happened because someone from the human cast picked it up and noticed it could be funny. Eight years ago, and because of a bug on my side, A.L.Ex would sometimes generate computer code. So we all adopted the word BR to punctuate our sentences, like “See you tomorrow! BR.”. Last year in Edinburgh, we had an improv scene where Holly Mallett (who sings and composes musicals herself) was the Cyborg. When A.L.Ex started sending her lines that looked like a song, Holly picked up on the rhymes and rhythmic structure and burst into a song, interactively written in real time.
You’ve worked in both cutting-edge science and live theatre. What connects those two worlds for you?
Failure—and a “maker” mindset—are what is common to both improvisation and to the scientific method, and I see so many parallels between the two. Improv is all about making offers to your stage partner, knowing that only some of them will work and be funny, and about taking and accepting that risk. Science is all about collaborating, trying new ideas, knowing that most experiments will fail. Like good scientists, improvisers learn from their mistakes over their training, and they also make hypotheses at the start of each scene (Who? What? Where? What is the game of the scene?). Like good theatre people, scientists often improvise scrappy new technical solutions with limited budgets and time.
What’s the biggest misconception people have about AI in creative work?
I would love to quote poet and software engineer Allison Parrish (www.decontextualize.com) who presented and disputed two fallacies about AI in artistic work.
The first fallacy is that “creative labour can be automated” and that AI can in any way replace artists. Like every Edinburgh punter knows, we are here to experience the art made by vulnerable human beings, and to connect to their lived human experience, which simply cannot be automated, no matter how advanced AI technology will become.
The second fallacy is that “writing a computer program to generate aesthetic artifacts is not in itself a creative process” (assuming, of course, that one does more than just type a prompt into a generative AI system). I love creative coding and building robots for performance, because these are strange tools whose default behaviour is to disobey your instructions—until you realise you made a bug in your code that you need to fix. Theatre can be a great playground for pioneering technologies: lights, projections, immersive installations, and now robots. It is even more appropriate that the word “robot” was actually first coined on the theatre stage in the theatre play “R.U.R. – Rossum’s Universal Robots”, which was written over 100 years ago by Czech playwright Karel Čapek.