Gupta says, “Every AI for everyone is like our tagline.” We have organized all AI models that we can find today. “The UP website encourages developers that if they want to add their language or image model to the options, they do not encourage them to reach it, and they do not currently call it.
Each time someone uses UPP, he is contributing to the head of two chatboat models, and sometimes a reward for providing his opinion and selecting a winning response. Basically, this is a user survey that has a fun game. (Website is A lot of Emoji.)
He sees the data trade in this situation more clearly than past consumer apps in this situation, such as Twitter – which he is in a hurry to tell me that he was a 27th employee and is now one of the company’s Coofer, Baz Stone, as a supporter. “This is a little leave from previous consumer companies,” he says. “You provide feedback data, this data is being used anonymously and will be sent to the model builders.”
Which we have the real amount of money: Selling human beings to AI companies that intensely want to improve your model more data.
“The human diagnosis of the crowd is what we are doing here,” says Gupta. It is estimated that the amount of consumers will increase some cups of coffee in a month. However, such data labeling, which is often called kick -learning with human impression in the AI industry, is extremely valuable to companies as they release repentance models and fix outpots. It costs more than a coffee cup of coffee in all of San Francisco.
The main rival of the UP is a website called Lemrina, which is very popular in AI internal to get feedback on new models and if a new release is reached to the top of the pack. Whenever a powerful model is included in Lemarina, it often sticks rumors about which the big company is trying to test its new release in Steelth.
“This is a two -sided product that has network effects that help model builders, Gupta says.” “And hopefully the model maker is improving the models and sending them back to the consumers.” He shows me the beta version of the Leader Board of the UP, which runs directly today and includes more granular data as well as the overall rating of models. The ranking can be filtered with the extent to which a model performs better with specific settlement information that performs with consumers like their age, or in a particular instant category, such as combined with specific settlement information with health care questions.
Near the end of our conversation, Gupta reveals the theory of artificial general intelligence, such as supernetilization, human beings. He says, “These models are being made for human consumers at the end of the day, at least for the near future.” Among the people working in AI companies, this is a generally common belief and marketing point, even though many researchers are still raising the question of whether the basic technology behind big language models can ever produce AGI.
Gupta UP users, who may be anxious about the future of humanity, want to imagine themselves to actively create these algorithms and improve their quality. “It’s better than free, because you are doing this great job for the future of AI.” “Now, some people want to know this, and others just want the best answers.”
And even more consumers may want extra cash and spend a few hours spending their opinions during their chat boot conversation. I mean, $ 50 is $ 50.
