In 2018, researchers at MIT demonstrated a device enabling people to search Google with their brains.
Two years prior, the Neuralink Corporation launched efforts to develop implantable brain-computer interfaces, microchips enabling paralyzed people to use mobile technology via neural signals.
But even before that, in 2009, Vauhini Vara began writing a novel which features a device allowing a person to connect their brain directly to the internet.
As an emerging technology reporter and author, Vara is engrossed by humans’ continual innovation. Her book, The Immortal King Rao, explores artificial intelligence at a crossroad with society.
“I find it fascinating that humans are always curious and always trying to invent new things, for better or worse,” she said.
“The spirit of innovation never stops,” she wrote in The Immortal King Rao, her debut novel and a finalist for the 2023 Pulitzer Prize in Fiction.
That spirit can bring the world “fascinating things” like ChatGPT, Vara said, but “there are two sides to that coin.” The same technologies might also infringe on copyrights and threaten jobs. At 10:45 a.m. today in the Amphitheater, Vara will continue the week’s Chautauqua Lecture Series theme, “A Life of Literature,” centering her lecture on the opportunities and implications of artificial intelligence.
Vara first encountered AI as a reporter in the mid 2000s, around the time startup companies such as Facebook and Google were on the rise — an “interesting time for tech,” she said. Vara found the tech industry intriguing once she started to report on it, and has been doing so ever since.
In her 2021 essay “Ghosts” for The Believer, Vara experimented with using AI in her writing.
She reached out to Sam Altman, cofounder of OpenAI, to try out a new tool, GPT-3, a language prediction model using AI to write based on inputed words.
Vara began by “plugging in random sentences and seeing what it would spit out,” she said. Through that process, it occurred to her that the technology could be useful for helping people “write about things that are otherwise really difficult to write about.”
For Vara, the loss of her sister to cancer has been a hard topic to communicate, so she collaborated with the AI to write “Ghosts,” resulting in an unpacking and thorough expression of her grief and emotions.
Although artificial intelligence can be useful, Vara admits the ability of OpenAI to stand in for what “really should involve some human effort” — helping humans express difficult things — is part of what makes these technologies “insidious.” Her essay plays with that idea, manifesting what it looks like for AI to stand in for human contemplation.
“The technology helped me unlock my ability” to describe her sister, Vara said. “At the same time, careful readers will also be aware that the essay plays out, on a meta level, this idea of what the technology is doing to attract us.”
Vara said she hasn’t come to a definitive conclusion about the appropriate place for AI in the literary world.
“For me as a writer, the purpose of writing is to express something of my own singular consciousness,” she said. “That is not something that AI can do.”
Still, Vara said, using an AI language model “somehow helped to draw out something of my own consciousness, my own ability to express myself. The fact that that happened, and the fact that the AI produced some really brilliant lines as part of that process, makes it hard for me to totally dismiss AI.”
At several moments in “Ghosts,” Vara said she felt GPT-3 was able to write “something really beautiful, something that moved me,” Vara said. She feels many people are quick to deny AI and think that it will never write like humans, that “it’s not a threat because it’s not human. It (won’t) do anything close to what we can. Some of the lines that the AI wrote call that into question.”
When Vara looks to the future and imagines what neural networks might become as the technology advances, she recognizes “the one thing AI is not, is a representation of individual human perspective.” AI writing may lack emotional awareness and cohesiveness now, but many of those flaws “are technical problems that can be solved” with time. A moment in the future may come when the imperfections are smoothed over.
But right now, Vara said, human literature is unique in expressing the singular consciousness of the author. AI language models serve a different function — less a comprehension of human writing than a synthesis of available samples on the internet, which she noted “may have its own merits.”
“We, as humans, have (a) strong need for connection to one another,” Vara said. “There’s something really potent about the idea of a technology that can connect us all, something moving about it.”