What if the answer to your next chatbot question came from a stranger instead of an artificial intelligence system? That is the idea behind ChatTJB, an unusual chat service that replaces AI-generated responses with answers written by real people. Created by Tucker Bryant, a former Google employee, the platform was launched in April as an art project but quickly became an unexpected internet sensation.
The concept arrives at a time when people increasingly rely on AI assistants to make decisions, find information, and even provide personal advice. ChatTJB asks users to pause and consider a simple question: Do we always need a machine to think for us?
What is ChatTJB?
ChatTJB looks and feels like a conventional AI chatbot.
Users type a question into a text box and submit it. The crucial difference comes after that: there is no AI model generating the response.
Instead, a human reads the question and writes an answer.
The people responding are described by the project as “average individuals”. They are not presented as an army of artificial intelligence experts. The entire premise is built around getting another human being to think about a question and respond.
That can mean the answer takes considerably longer than users expect from ChatGPT or other AI assistants.
But the delay is part of the experiment.
Why did Tucker Bryant create ChatTJB?
Bryant’s idea reportedly came from an ordinary interaction with an AI chatbot.
He found himself asking a chatbot for advice about what sleeves to wear on an 18-degree Celsius day in a city where he had lived for seven years.
The question made him think about how easily people can outsource even trivial decisions to machines.
That led Bryant to explore what he describes as “cognitive surrender”, the idea that people can gradually hand over parts of their own decision-making process to AI.
The project therefore isn’t simply an anti-AI chatbot.
It is an experiment about what happens when people become accustomed to asking machines for answers that they might otherwise work out themselves.
What is “cognitive surrender”?
The term points to a broader concern surrounding increasingly capable AI assistants.
AI can be useful when it helps people process information, identify options, or complete difficult tasks. But there is a difference between using a tool to assist with thinking and allowing the tool to replace the thinking altogether.
Research cited in the source material from Wharton professors Steven D. Shaw and Gideon Nave examined how people performed with AI assistance.
According to the research cited by NewsBytes, people using AI could perform better when the AI’s assistance was correct. However, when the AI provided incorrect advice, their accuracy fell by 15 percentage points compared with people who did not receive AI assistance.
That finding illustrates a potential problem with overreliance: an AI system can be useful while still making people less likely to question a confident but incorrect answer.
ChatTJB takes the idea in the opposite direction. Instead of giving users a faster machine-generated response, it deliberately puts another human being back into the loop.
How did ChatTJB suddenly become popular?
The project gained significant attention after a billboard advertising ChatTJB appeared in San Francisco.
The publicity produced a surge of people visiting the service and submitting questions.
According to the source material, ChatTJB received more than 100,000 prompts after the billboard went up. At one point, users were reportedly submitting more than 5,000 prompts every hour.
For a project designed around individual humans answering questions, that created an obvious problem.
There simply were not enough people available to respond immediately.
Bryant began recruiting volunteers to help handle the growing number of questions.
More than 10,000 people have reportedly applied to assist with the service, while the volunteer waitlist has been increasing by roughly 1,000 applicants a day.
Why are ChatTJB’s answers so slow?
Speed is one of the biggest differences between ChatTJB and conventional AI assistants.
An AI model can process a prompt and produce an answer in seconds because the response is generated computationally.
ChatTJB depends on people.
A volunteer has to see the question, understand it, decide what to say and then write the response. If thousands of people are waiting for answers, the queue can become substantial.
That means ChatTJB can sometimes take days to respond.
The source material gives image generation as one example. When a user asked ChatTJB to create an image, the service did not instantly generate one. Instead, it explained that thousands of people were submitting questions and that the volunteer response could take days.
The inconvenience is almost the point.
ChatTJB turns the instant-answer expectation of the AI era into a slow, human process.
What happens when people ask ChatTJB personal questions?
The project’s popularity has also created an unexpected consequence: people are sending deeply personal questions.
That raises an interesting distinction between human and AI assistance.
An AI assistant can respond without embarrassment, fatigue or judgment, but it also lacks human experience in the ordinary sense. A person responding through ChatTJB, meanwhile, brings their own experiences and perspective to the interaction.
That does not necessarily make a human response more accurate.
It does, however, make the exchange fundamentally different.
A stranger may respond with empathy, skepticism, humor or a perspective shaped by their own life. The unpredictability is part of what makes the project different from an automated chatbot.
Users are effectively asking the internet to find another person willing to think about their problem.
Is ChatTJB really an alternative to AI?
Not in the conventional sense.
ChatTJB is better understood as a social and artistic experiment that deliberately removes automation from the chatbot experience.
It does not attempt to outperform AI models on mathematics, coding, research or information retrieval.
Instead, it challenges the assumption that every question should have an immediate machine-generated answer.
That distinction matters because the platform’s usefulness depends heavily on the question.
For a complicated technical problem requiring verified information, an AI system or human expert may be more appropriate than a random volunteer.
For a question about perspective, everyday decisions or curiosity, however, a human response can provide something an algorithm cannot easily replicate: another person’s subjective experience.
Why does the project matter in the age of AI?
ChatTJB arrives during a period when AI assistants are becoming increasingly embedded in everyday life.
People use AI to draft emails, plan trips, write code, summarize documents, brainstorm ideas and make decisions.
The convenience is obvious.
But convenience can change behavior.
If someone asks an AI system every time they need to make a small decision, they may gradually become less comfortable making those decisions independently.
That is the question Bryant appears to be putting at the center of ChatTJB.
The service does not ask whether AI is good or bad. It asks what happens when humans become so accustomed to automated answers that they stop noticing how often they use them.
What happens to ChatTJB next?
The project’s unexpected popularity has created a problem that Bryant did not originally have to solve.
ChatTJB began as an art project. Its growth has turned it into a service requiring volunteers, infrastructure and ongoing management.
Bryant has said he wants the project to exist long term, but maintaining it is not currently his primary source of income.
That leaves ChatTJB facing a question that many experimental internet projects eventually encounter: Can an idea designed to make a point survive once thousands of people start relying on it?
For now, its growing volunteer base suggests that there is considerable interest in the experiment.
ChatTJB’s bigger question for AI users
The most interesting part of ChatTJB may not be whether a stranger can provide a better answer than an AI model.
It is what the project forces users to notice.
AI assistants have made waiting for an answer feel unnecessary. ChatTJB reverses that expectation. The user submits a question, waits for another person to read it and accepts that the response may arrive much later.
That friction is intentional.
It reminds users that answering a question is not always the same as generating text. Sometimes the value lies in another human being taking the time to consider what was asked.
ChatTJB may never replace mainstream AI assistants, and it does not need to.
Its more unusual contribution is to hold up a mirror to the chatbot era: when a machine can answer almost anything instantly, perhaps the more interesting question is whether we should ask it everything in the first place.