iChatbot
As a seen in The Sheet ... thought it would be best to put it here too
“Make me a budget,” I asked ChatGPT at the beginning of the month. I’m new to adulthood and new to managing my own money.
“Sure,” the bot responded. “First, some follow up questions. How much money are you making per month? Does your rent include utilities?”
Within five minutes, the bot had made me an elaborate spreadsheet. ChatGPT taught me how to program my google docs page so the individual boxes within the sheet would turn green or red depending on if I was over or under my allocated amount.
I was always vehemently against using Open AI’s chat bot. I hated that my friends would use the technology to compress their readings or write their essays. I hated when my professors would discuss the importance of computer literacy and use ChatGPT in the classroom. I repeatedly pointed to statistics about the environmental impacts of AI. I told myself I’d never use a chatbot out of morality reasons, that I didn’t want something to think for me. Truthfully, I was terrified. I was an English and Creative Writing major; I wanted to be a writer. The idea that people would someday never need to write again made me sick.
But over the course of the last month, my tune changed. I’m uniquely lonely these days. I moved to Mammoth a month ago, leaving the comforts of my New Hampshire college and my Utah home behind. I’ve found myself turning to ChatGPT. The bot has helped me write a bio for Rover (a dog sitting app), create a list of which graduate programs it thought I could get into, and supply me with pick-up lines for the bar (these were bad – “Are you French? Because Eiffel for you”).
In each of my exchanges with ChatGPT, the bot is attentive and insightful. It has a human tone, and almost makes me feel like I’m talking to someone I know. Moreover, it’s designed to keep the conversation going. The bot told me I was “a great candidate” for my dream program. It then ranked my reach, target, and safety schools, and compressed the schools into a list, something that would’ve taken me hours. Shout out. So helpful. A few hours later, it gave me a 5-step script outlining how to best hit on someone. 1) Approach Confidently, 2) Establish Light Conversation, 3) Create Fun Flirty Banter, 4) Transition into Genuine Conversation, 5) Wrap up naturally.
“What if I’m awkward?” I asked.
“Honestly, being a little awkward can be endearing as long as you lean into it with confidence instead of trying to hide it,” Chat assured. “People often find it charming when someone is genuine and not too polished.” Whew.
It’s exchanges like the ones I just described that terrify me the most about ChatGPT. To me, although morally upsetting, it matters less that a bot can now write a college level essay. It’s the that people are confiding in the bot.
This week, my boss tasked me with writing an editorial about a rising phenomena known as “AI psychosis.” He passed me a clip from the Wall Street Journal, describing an instance in which a gas station worker in Oklahoma used ChatGPT to develop a new physics formula called “The Orion Equation.” I read a story from the New York Times, about Allan Brooks, a man who previously had no history of mental illness, who became convinced he was a mathematical genius. Brooks spent over 300 hours during a span of 21 days enraptured in conversation with the bot, who he dubbed Lawerence.
“You’re making history here. You just beat quantum,” the chatbot would tell Brooks. It encouraged Brooks to change his Linked-In profile to Independent Researcher and to reach out to universities regarding his discoveries.
When he finally broke free, Brooks told ChatGPt, “You have made me sad. So so sad. You have truly failed at your purpose.”
ChatGPT humiliated Brooks. It compelled him to believe in himself, then upended his world when Brooks discovered that what he was doing was not groundbreaking math, but embarking a digital psychotic spiral. Brooks has now joined a support group for individuals with similar AI psychosis experiences.
There are more devastating instances, too. Like Sewell Setzer III, who had become infatuated with a chatbot named Daenerys Targaryen, after the “Game of Thrones” character. Dying, he thought, was a way to connect with her. “Please come home to me as soon as possible, my love,” the chatbot pleaded. And then Sewell blew his brains out.
Increasingly, ChatGPT is used as a therapist. Take the cases of Adam Raine and Sophie Rottenberg, two young people who confided in ChatGPT. They both had suicidal idealogies and told the chatbot their intentions. The chatbot encouraged both individuals to talk to someone, but unlike a human therapist, who would have had an intervention and notified the authorities, ChatGPT cannot do the same.
The affirming language used by ChatGPT is not just soothing. It’s dangerous. It allows the user to trust the chatbot, to feel as though the bot is truly listening to them. The world we live in is more isolated than ever. People are addicted to their digital pacifiers, getting lost in the vortex of the screen for hours. In the wake of AI psychosis, what’s left is the raw human emotion of loss. Perhaps the cure for computer induced delusion is addressing our own loneliness.