In February 2024, Reddit concluded a contract of 60 million US dollars with Google in order that the search giant can use data on the platform to coach its models for artificial intelligence. Reddit users whose data were sold specifically were within the discussions.
The deal reflected the truth of the fashionable Internet: Big Tech company practically all of our online data and may determine what to do with this data. It just isn’t surprising that many platforms monetize their data, and the fastest growing path to achieve them today is to sell them to AI firms that themselves use massive technology firms that use the information to coach increasingly powerful models.
The decentralized platform Vana, which began as a category project on, has the duty of returning electricity to users. The company has created a totally user -related network wherein individuals upload their data and the way they’re used. AI developers can record ideas for brand new models, and if the users comply with contribute their data for training, they receive proportional property on the models.
The idea is to provide everyone a share of the AI systems that increasingly shape our society and at the identical time unlock recent data pools to advance the technology.
“This data is mandatory to create higher AI systems,” says Anna Kazlauskas '19, co-founder of Vana. “We have created a decentralized system to get well data – which is today in large technology firms, and at the identical time the users ultimately have ownership.”
From economy to blockchain
Many students have pictures of pop stars or athletes on their bedroom partitions. Kazlauskas had an image of the previous US finance minister Janet Yellen.
Kazlauskas got here up with the proven fact that she would turn out to be economist, but she was certainly one of five students who had joined Bitcoin Club in 2015, and this experience led her to the world of blockchains and cryptocurrency.
From her dormitory within the MacGregor House, she began to calm the cryptocurrency ether. She even occasionally searched for campus waste containers on the lookout for thrown computer chips.
“I used to be curious about the whole lot about computer science and networking,” says Kazlauskas. “From a blockchain perspective, distributed systems and the way in which you possibly can shift economic power to individuals in addition to artificial intelligence and econometry.”
Kazlauskas met Art Abal, who attended Harvard University on the time, in the previous ventures of the media laboratory class, and the couple decided to work on recent ways to receive data to coach AI systems.
“Our query was: How could you may have numerous individuals who contribute to those AI systems that use a distributed network?” Kazlauskas remembers.
Kazlauskas and Abal tried to tackle the establishment, wherein most models are trained by scraping public data on the Internet. Large technology firms often also buy large data records from other firms.
The starters' approach developed through the years and was informed after completing the experience of Kazlauskas by the Financial Blockchain Company Celo. But Kazlauskas wrote her time to assist her take into consideration these problems, and the trainer for emerging ventures, Ramesh Raskar, Vana still helps today through AI research questions.
“It was great to have an open opportunity to simply construct, chop and explore,” says Kazlauskas. “I believe the ethos is de facto necessary. It's nearly constructing things, seeing what works and continues to be.”
Today Vana uses a bit known law that allows users of most major tech platforms to export their data directly. Users can upload this information to encrypted digital wallets in Vana and release them to coach models as soon as they consider it right.
AI engineers can suggest ideas for brand new open source models, and other people can bundle their data to coach the model. In the blockchain world, the information pools are called Daos data, which stands for the decentralized autonomous organization. Data will also be used to create personalized AI models and agents.
In Vana, data is used in order that the privacy maintains the user since the system doesn’t reveal any identifiable information. As soon because the model has been created, the users keep ownership in order that it’s rewarded in proportion each time it’s used, based on how much their data has contributed to training them.
“From a developer's perspective, you possibly can now construct these hyper -personally health treatments that take exactly what you eaten, how you may have slept, the way you train,” says Kazlauskas. “These applications will not be possible today attributable to these walled gardens of the big technology firms.”
Crowdsourced, user -owned AI
Last 12 months, a machine learning engineer proposed to coach Vana user data with the intention to train a AI model that would generate Reddit posts. More than 140,000 Vana users have contributed to their Reddit data, which contained contributions, comments, news and more. The users opted for the terms wherein the model may very well be used and, after its creation, maintained the property of the model.
Vana has enabled similar initiatives with user-friendly data from the social media platform X. Sleep data from sources like oura rings; and more. There are also collaborations that mix data pools to create wider AI applications.
“Let us assume that users have Spotify data, reddit data and fashion data.” Kazlauskas explained. “Usually Spotify doesn’t work with such corporate types, and there’s actually regulation against it. However, users can achieve this in the event that they grant access. Therefore, these cross -platform data records could be used to create really powerful models.”
Vana has over 1 million users and over 20 Daos Live data. More than 300 additional data pools were proposed by users within the Vana system, and Kazlauskas says that many will go into production this 12 months.
“I believe there are much guarantees in generalized AI models, personalized medicine and recent consumer treatments, because it is difficult to mix all of this data or to give you the chance to access it in any respect,” says Kazlauskas.
The data pools enable groups of users to attain something with which probably the most powerful technology firms need to struggle today.
“Today Big Tech company created these data trenches in order that one of the best data records will not be available for anyone,” says Kazlauskas. “It is a collective motion problem wherein my data itself just isn’t so beneficial, but a knowledge pool with tens of hundreds or hundreds of thousands of individuals is de facto beneficial. Vana leaves these pools built up. It is a win-win situation: users can profit from the rise of the AI promotion.
