I Am Alan Turing: An Opera By Humans and AI
Long before the AI craze that swept the world in 2022, British mathematician Alan Turing envisioned—as early as the 1950s—a machine that could think and learn like humans.
Devising an opera about Turing was an idea that had already been in the works when a unique opportunity presented itself. What better way to pay tribute to the visionary than to make the very machine he once foresaw as a creative partner in this production?
In the thick of the pandemic, with the world in lockdown, a group of talented creatives got together to make this artificially intelligent opera a reality: I AM ALAN TURING.
The Imitation Game
“A computer would deserve to be called intelligent if it could deceive a human into believing that it was human.” – Alan Turing in his paper Computing Machinery and Intelligence.
Posed as the Imitation Game, this was one of the first formal definitions of machine intelligence that eventually came to be known as the “Turing Test”—a standard benchmark for AI models today.
In the same paper, Turing also suggested that instead of programming every rule into a machine and trying to imitate a fully developed adult brain, the goal should be to make a “child machine,” which, like a child’s brain, could develop through learning, training, and feedback.
Sounds familiar? This is the basis of modern GPT models, whose answers get refined as they learn through user experiences and gather more data with every passing day.
Reimagining an Epic Life
“We've been using the word opera to denote a sense of scale,” explains sound designer and composer Frederick Kennedy, who is an assistant professor in the Department of Theatre Arts & Dance and one of 14 collaborators involved with the I AM ALAN TURING project.
Turing is famously known for helping decrypt German cipher messages during World War II by cracking the Enigma machine. The devastating, “rather operatic” arc of Turing’s life was that he was prosecuted for being a homosexual. Turing was forced to undergo chemical castration as an alternative to prison before he eventually took his own life two years later, in 1954.
“It’s less about singing in a traditional operatic style and more about this idea of a complete large-scale artwork to depict a very epic, noble, brilliant, tragic life,” says Kennedy.
The devised multimedia opera now stands at an hour long, broken into several acts. (“Devised” refers to a collaborative creation process where a team develops the piece improvisationally, rather than a single playwright working alone.)
From a musical perspective, I AM ALAN TURING draws from a variety of genres, ranging from contemporary pop and experimental electronic music to classically sung pieces. “[It’s] kind of like a collage, grabbing bits and pieces of ideas and piecing them together to make something that's new and fresh and interesting,” says Kennedy.
In a way, he reflects, it almost embodies the spirit of how large language models (LLMs) work—pulling from a large and eclectic palette of sources to generate something novel.
The Newborn Brain: GPT
In 2019, acclaimed composer and professor at Yale University, Matthew Suttor, gave a lecture on creativity to the Department of Genetics at Yale School of Medicine. Afterward, a colleague suggested he meet “some guys in San Francisco.” It turned out to be OpenAI. Intrigued by Suttor’s project, the startup wanted to collaborate with creative practitioners before launching its AI models—something it had never done before.
OpenAI advised the group on fine-tuning an instance of GPT-2 that ran on a server at Yale’s Digital Humanities Lab. The team built their own Application Programming Interface (API) to interact with GPT-2 and trained it on papers written by Turing, books he was known to have read as a child, and plays by George Bernard Shaw, whom Turing admired.
“The idea was to essentially create an avatar of Alan Turing based on all of the language that we know he had consumed, said, or written,” says Kennedy, who joined the project the following year after being brought on by Suttor, his mentor during his graduate years at Yale.
Once the model had been sufficiently trained, the team began to pose questions to it. One of the first questions asked was “Who are you?” to which the model replied: “I am Alan Turing. I am here to tell you you are not alone in the world. Here is my story. It is a story about the rise of the computer. It is a story about the rise of the government. It is a story about the rise.”
A goosebump-inducing response, Kennedy emphasizes that this was before any AI model was publicly available. “It wasn't in the news, it wasn't in the popular consciousness, and so the sort of uncanny valley, strange and magical experience of having this technological interaction was just really shocking and profound.”
They asked other questions like “If you wrote a song about an Enigma machine, what would the lyrics be?” or “Can machines think?” Being an earlier model that was mostly unadulterated, most of the answers were chaotic and peculiar, almost like poetry sprinkled with the dreamlike quality of making no sense.
It was this iterative process of re-feeding prompts, rewriting, and piecing together different materials that finally shaped the libretto of the opera.
Bringing Out the Inner Child
In a way, the GPT model was an object of curiosity for the artists, much like a plaything is for a child. “Even when these strange answers come back, it's like, ‘Oh, that’s weird, what if we take that and connect that with this?’ And at a certain point, you're just playing with it like this amazing toy, which in this context, is really what it is,” says Kennedy.
It was the combination of having a sense of whimsy backed by years of scholarly training that made the human and machine collaboration so fruitful, teasing out all creative possibilities.
For many of us today, AI models have almost started playing the convenient role of traditional search engines, giving us quick, concise answers from a collection of sources. Using AI in I AM ALAN TURING, however, has not been a shortcut whatsoever.
“The point of the tool isn't to simplify work for anyone on this team. The point of the tool is to help us rework and reimagine, and take us in new and unexpected directions,” clarifies Kennedy.
Surprisingly, using AI was not a time-saver for the team. “We’ve probably spent two times more energy writing this thing than we would have if we just sat down and said, let’s write an opera.”
Is Using AI Creatively Ethical?
Netizens have been very critical of the usage of AI for creative purposes. Not only is there concern about copyright issues, but also about AI replacing artists.
According to Kennedy, pretending this emerging technology doesn’t exist isn’t particularly helpful to anyone. It’s about who is using it and how that makes the difference.
“I think we're in a kind of challenging moment right now, where this tool is being shipped out to all kinds of students who see it as a way to offload work. But they haven't yet developed the critical faculties to engage with whether the work is good or meaningful, or how to take it and shape it into something more,” rationalizes Kennedy. “So, if you're already a fully formed artist, person, writer, whatever it is, it can be a really powerful editorial tool.”
While acknowledging there are many serious concerns about the emerging technology, he is also skeptical about the argument that AI is an existential threat to artists’ livelihood, mainly because there is one thing humans will always have that machines cannot replace—physical connection.
“One thing about the theatre, and live performance in general, is it's about human beings sharing physical space together. It's not digitally mediated, so if I'm sitting in the audience and there's a performer or a singer who's 15 or 20 or 40 feet away from me, there's a literal physical connection that a computer just can't replace, you know?”
In some strange ways, the very technology people fear will drive us apart could ultimately bring us together in a more intimate way.
What role do the liberal arts play in understanding and scrutinizing artificial intelligence?
Advances in AI are social, cultural, and ethical. As these new technologies reshape industries and influence our daily lives, they also raise important questions about the relationship between people and technology. That's why we need the liberal arts. The liberal arts provide the context to grapple with these questions through a humanity-centered perspective.
This story was written by Anushka Raychaudhuri, an undergraduate student in CLA.
