This chatbot tool pays users $ 50 per month for their feedback on models

“” Every A for all “is a kind of our slogan,” says Gupta. “We organized all the models that we can find today.” The Yupp website encourages developers to reach if they want their language or image model added to the options. At the moment it has no offer with the AI ​​models manufacturers and provides these answers by making bees calls.

Whenever someone uses Yupp, participate in a head -to -head comparison of two chatbot models and sometimes receiving a reward for providing their feedback and choose a winning response. Basically, it is a user -disguised user survey. (The website has many of emoji.)

He sees the data exchanged in this situation for users as more explicit than past consumer apps, such as Twitter, who is ready to tell me that he was the 27th employee and now he has one of the co -founders of that company, Biz Stone, as one of his supporters. “This is a bit of a departure compared to previous consumer companies,” he says. “Provide feedback data, that the data will be used anonymously and sent to models manufacturers.”

Which takes us to where the real money is found: to sell human feedback to artificial intelligence companies that desperately want more data to develop their models.

“The human assessments of Crowdsourcing are what we are doing here,” says Gupta. Estimates that the amount of cash that users can add enough to some cups of coffee per month. However, this type of data labeling, often called Reinforcement learning with human feedback In the artificial intelligence sector, it is extremely precious for companies as they release iterative models and develop the results. It is much more worth the cup of foolish coffee in San Francisco.

Yupp’s main competitor is a website called LMANAWhich is quite popular among artificial intelligence professionals to receive feedback on new models and boast if a new version rises to the top of the package. Whenever a powerful model in Lmarena is added, it often feeds the voices on which large company is trying to test its new invisible liberation.

“This is a two -sided product with consumer network effects that help models manufacturers,” says Gupta. “And the models manufacturers are hoped, they are improving the models and are sending them to consumers.” It shows me a beta version of the Yupp ranking, which is live today and includes a general classification of the models together with multiple granular data. The rankings can be filtered by the way a model behaves with specific demographic information that users share during the registration process, such as their age or in a particular prompt category, such as questions relating to health care.

Towards the end of our conversation, GUPTA makes artificial general intelligence appear-the theory of superintendent algorithms, similar to man-like imminent technology. “These models were built for human users at the end of the day, at least for the near future,” he says. It is a fairly common belief and a marketing point, among the people who work in companies AI, although many researchers still wonder if the technology underlying at the base of large models will ever be able to produce agi.

GUPTA wants Yupp users, who could be anxious for the future of humanity, imagine themselves as actively modeling these algorithms and improve their quality. “It is better than free, because you are doing this fantastic thing for the future of Ai,” he says. “Now, some people would like to know and others only want the best answers.”

And even more users could only want extra money and be willing to spend a few hours to give feedback during their chatbot conversations. I mean, $ 50 are $ 50.