Achieving AI’s full potential depends on data collaboration. But with data more decentralized than ever, a new approach is essential.
In this episode, Matt Karasick, SVP of Product at LiveRamp, explains how clean room technology brings organizations together. Whether it’s training fraud detection models for highly regulated industries like banking or enabling collaboration across multiple clouds, clean rooms are unlocking new use cases.
Watch to learn how clean rooms make it easier for companies to develop AI models securely.
What do you do when you want to collaborate with a company who doesn’t trust anyone? I’m Matt Karasick, Vice President of Product for Insights at LiveRamp and here’s the deal.
Clean rooms have emerged as the way for people to be able to collaborate using each other’s data models or code to produce insights and outcomes. And at the core of clean rooms, it is to eliminate the need for partners to trust each other, to trust their technology providers, to trust their clouds. Let’s take an example of five banks wanting to collaborate. Now, both because it’s regulated and because they’re competitive, banks can’t just go and start shipping each other customer data, transaction data. It’s not even if they all shook hands and said, “We’ll only use it for X, Y and Z.” That would take trust. That would mean that the attorneys would have to put a thousand pages of rules in place to say, “We will do this, we won’t do this.” And at the end of the day, there are certain companies who are simply not going to rely on trust.
Rather they’re looking for technical guarantees to say only these things can occur and nothing else. So take the example of these banks wanting to build better fraud detection models. That is mutually beneficial to all of the banks, to all of us consumers who use these banks. Clean rooms are emerging as a way to remove the need for trust there to make it so that they can train a new model using each other’s data, models and code. They can reinforce this model and continue to improve it without the possibility of any other outcome occurring. And because of the lack of need for this trust, they’re able to implement and execute these use cases at scale.
It has changed the game. It also needed to change the game. As the adoption of cloud has proliferated, data and models in code have become decentralized, they have become distributed. And this is the way in the world. There’s almost no customer out there who is not using more than one cloud at once. And so given that there is more data than ever before, it is more distributed than ever before. The alternative is everyone having to trust each other, choose one cloud, one platform, and centralize all their data. And that’s never going to occur. Or they need a way to do it, where data stays where it is and it has all these protections without the need for this type of trust and these types of contracts. And so absolutely, this is the unlock to allow companies to use more data than ever before.
