New Delhi: What if, instead of asking ChatGPT a question, you asked a random stranger on the internet?
That is the premise behind ChatTJB, an unusual chat window which is powered by humans instead of AI. Its creator, San Francisco-based artist and former Google employee Tucker Bryant, started the project as an art experiment to find out how readily people turn to AI for answers.
The joke is hidden in the description. ChatTJB describes itself as being powered by “AI”, where AI stands for “Average Individual” and not artificial intelligence.
Unlike conventional chatbots, ChatTJB does not use a large language model to generate responses. Users submit questions through the website, and Bryant reads them and replies himself. The site’s own description says that a person named Tucker reads a question, thinks about it and responds “when awake and motivated”.
The questions can be as ordinary as how to cook an egg, what to wear or what to have for dinner. In one example reported about the project, users even asked for creative suggestions such as names for houseplants. The appeal is partly that the answers are imperfect — and deliberately so.
Bryant launched ChatTJB in April, but for several months it remained a relatively small internet curiosity. That changed after he spent $6,000 on a billboard in San Francisco on July 27 to advertise the service. The billboard quickly attracted attention online, sending traffic soaring.
At one point, ChatTJB was receiving around 5,000 requests an hour. The sudden popularity became difficult for one person to handle, forcing him to temporarily pause the service while he worked on safeguards and additional support.
A community effort
The project has since changed shape. Bryant says ChatTJB is being developed as a community effort, with approved volunteers to answer questions. About 10 volunteers have reportedly joined so far. The website now warns users that they are interacting with an actual human being.
In a message displayed on the site, Bryant says the unexpected response has made it impossible for him to continue running ChatTJB alone.
“This got more attention than I anticipated and it is no longer conceivable for me to keep up,” he writes. He says he wants the project to survive long-term but would need an operating partner or sponsorship to do so.
The experiment is not necessarily an argument against AI. Bryant says the project was intended as “a short-term art project to encourage people to remain curious about their and the world’s relationship to technology.”
There is also an unintended twist. Once users realise a real person is on the other side, conversations can become more personal and candid.
ChatTJB therefore turns the usual chatbot experience on its head: instead of a machine trying to sound human, humans are now trying to interact with other humans.
(Edited by Aakriti Handa)