A San Francisco billboard has been running since July 27, advertising a chatbot that is actually a 32-year-old man answering questions in real time.
The project is ChatTJB, created by Tucker Bryant, a former Google project manager and public speaker. The advertisement claims it is the leading chat interface powered by AI. In this instance, the acronym stands for average individual.
Bryant did not train a large language model to mimic his writing style. He sits at a desk and types responses himself. Requests for images of a skunk wearing sunglasses result in hand-drawn sketches. Questions about frying an egg receive suggestions that may or may not work. When WIRED asked how to become a morning person, the site noted that thousands of users were currently writing in, though no reply had arrived by the time of publication.
Bryant launched the service in April, but it gained traction only after the $6,000 billboard appeared. Coverage followed quickly. As of August 6, the project has handled over 30,000 queries. At its peak, Bryant was processing about 5,000 prompts every hour.
He describes ChatTJB primarily as an art project rather than a social experiment or a novelty act.
The work offers commentary on a strange moment we are currently living through, Bryant said. He does not view himself as anti-AI. He used the tool Lovable to design the website, yet he remains clear-eyed about the risks associated with the technology.
He worries about cognitive surrender, a term he uses to describe accepting AI outputs with little critical thinking or scrutiny. The concept made him reflect on his own interactions with large language models. He recalled asking an AI whether he should wear short or long sleeves on a 65-degree day in San Francisco.
He noted he had lived in the city for seven years and knew how to handle the weather. Still, it felt easier to defer to an AI for advice on that simple matter.
That feeling made him question his own role in the process.
Recent drawings include a frog eating spaghetti, San Francisco mayor David Lurie depicted as Godzilla, and two black cats eating grass on a table. The written requests have been equally diverse.
Some users asked for birthday party ideas when they did not want to host a celebration. Others sought dinner recommendations or advice on using ingredients in their refrigerator.
Many people initially treated the service as a joke. Once they realised they could have an actual conversation, the nature of the shared content shifted. Users began offering more personal and sincere details.
Bryant admits he is not a skilled artist and his advice is no better than that of any other person. On YouTube, he has called himself the worst LLM of all time. He believes users understand this limitation.
The value lies in having a human to bounce questions off of, rather than receiving expert counsel. Bryant found this dynamic very tender to see.
He expected the project to run for only a couple of days. Seeing user enthusiasm, he plans to continue. He has asked 10 volunteers to help, though they are a select group. Over 3,000 people applied for the role.
Superfans can support the work through ChatTJB Pro. Bryant labels this a worthless tier on the website. Payment provides the only benefit: the knowledge that one is supporting his art. So far, about four people have paid the $5 monthly fee. This has not covered the cost of the billboard, which will stay up until late August. Bryant stated he did not create ChatTJB to make money.
He wants people to reconsider their relationship with technology and the world’s relationship with it. Amidst companies selling world-changing technology and frightening reports of an AI bubble, he aims to bring some levity.
He hopes this approach makes a weird time sit a little bit lighter. He believes that is sometimes the best we can do.
What it means
The project highlights a growing reliance on automated systems for trivial decisions. By replacing the algorithm with a tired human, Bryant forces users to confront their own habits. The service proves that people often prefer human imperfection over perfect, instant answers.
