Sahara AI Ltd., the creator of a nascent decentralized artificial intelligence data network, today announced the launch of its new Data Services Platform, which allows everyone to contribute to the AI ecosystem by collecting, labeling, curating and refining datasets for model training.

The startup, which is backed by more than $43 million in funding, is building what it describes as a decentralized data marketplace for AI developers, who will be able to access large volumes of domain-specific data to train AI models for a variety of use cases.

In addition, its marketplace is designed to handle issues around copyright and privacy by allowing those who create the data to secure ownership of that information. This is done by providing attribution for their information that can be recognizably controlled even after it’s ingested into AI models during the training and fine-tuning processes, using decentralized blockchain technology.

The new Data Services Platform will allow anyone, including individuals, developers and large enterprises, to participate in a model that allows them to subscribe to data markets where contributors are fairly compensated for their efforts.

Sahara AI believes such a marketplace can be beneficial, pointing out that in the beginning, most AI models were trained on large swaths of publicly-available data from the internet, including public domain works. But much of this data comes from unattributed, copyrighted data sources, and this has caused a huge uproar. Artists, writers, news publishers, musicians, actors and other creators are demanding fair compensation when their data is used by AI models, and Sahara AI provides them with a way to tag and transparently control their ownership.

With Sahara AI’s model, data owners can be compensated each and every time an AI model is trained on, and then uses their information to generate something. Those who participate in its Data Services Platform will help to build the foundation of a more equitable and user-centric AI ecosystem, and be rewarded for doing so.

The startup argues that its data marketplace will also be necessary as AI evolves and becomes more advanced. Creating datasets used to be a fairly simple process that involved tagging objects and simple classifications, but more advanced AI models require much more nuanced datasets. They require sentiment annotation, multimodal data alignment and domain-specific datasets, and creating those is a complex task that involves thoughtfulness, expertise and high precision.

Sahara AI’s platform answers this need by leveraging a diverse, global pool of decentralized data labelers, and it has more than a few of them. More than 400,000 contributors have signed up to its waitlist to participate, and 10,000 have been selected to get the ball rolling. Some of its contributors include leading technology companies and innovators, such as Microsoft Corp., Amazon.com Inc., Snap Inc. and the Massachusetts Institute of Technology.

Those contributors will be able to select data tasks that span multiple industries, including the creator economy, financial services, sciences and more, with rewards distributed via a points system that measures the effort, accuracy and consistency of each one’s work.

Sahara AI’s marketplace has yet to launch, but it will do so soon. When it does, developers will be able to choose from a large selection of high-quality datasets and pretrained models, and also request that contributors create specific kinds of datasets, based on their model’s requirements.

Image: SiliconANGLE/Freepik AI