Home " News " MIT in practice: Creating insights in Dataville without sharing data

MIT in practice: Creating insights in Dataville without sharing data

Dataville is not an existing village. It is an AI solution developed by Lucidminds that allows organizations and companies to gain valuable insights from (potentially) sensitive data sets, without actually accessing that data. In short: no data sharing, but insights. And all that with great attention to control and privacy. A solution, therefore, that can eliminate many concerns.

"The developments in the field of AI are currently so rapid that there are concerns about it. And we share those concerns," said Bulent Ozel, co-founder and CEO of Lucidminds. "But that does not mean that the development of AI should be slowed down. On the contrary, we advocate driving the responsible use of AI as a technology. AI should be a tool for people and especially not a tool that will make decisions for us. So we should not let algorithms do all the work, but ensure that there is constant interaction between algorithms and human users. Always a human in the loop. And we have already developed several solutions for that."

Additional control options

One such solution is Dataville. That is a platform that allows data to be carefully and securely harnessed for AI applications. Says Ozel, "Because of concerns about privacy and security risks, there are several parties around the world, including technology giants such as Amazon, that are building similar solutions. But our solution consists of four different layers, including a layer that ensures that the data owner or data officer has additional control capabilities. And in that, we are unique."

Risk score for ensuring privacy

The premise of Dataville is that the data always stays where it is. So it is not shared. The insights derived from an analysis are. But that does not mean that privacy is thereby always guaranteed. For that, the algorithm models must also be known and transparent. But even then you're not there yet, because a party that has many analyses performed may at some point have accumulated so many insights that it may still be possible to discover sensitive information based on an analysis. To prevent that from happening, Dataville works with a risk score. If the risk is too high, the analysis cannot be performed. Unless there are reasons to do so. And that's where that extra control option comes in, with a human being making the final decision. There is also the possibility of refusing an analysis request when the risk score does not seem to warrant it. So the decision always lies with the human being.

Joint mission

The MIT budget awarded to Lucidminds helps continue to develop Dataville and make it available to an ever-growing group of companies and organizations. In addition, they are collaborating with several parties: Amsterdam City Council, Metabolic, VentureRock, HoTech and Leiden University. "Using the software we develop, we are building a bridge between theoretical science and practice," Ozel explains. "In doing so, we are trying to create a new mindset where people realize that in many cases it is indeed possible to use even sensitive data in a secure way for AI algorithms, without compromising privacy or security. At Dataville, that is surely our main mission."

UN development goals

And Lucidminds does all this from their office in Amsterdam's SDG House. They are in good company there, as the building houses a thriving community of entrepreneurs and organizations whose work focuses on the UN's development goals. Locally, they are doing their part to combat global problems such as extreme poverty, inequality, injustice and climate change. And so Lucidminds does that by deploying machine learning and creating simulations. Says Ozel, "Our efforts in that area are not aimed at replacing human intelligence, but rather at augmenting it so that we are better able to effectively address major societal challenges."

More information?

Visit the Lucidminds and Dataville websites. Or direct your questions to CEO Bulent Ozel.

Interested in the MIT scheme?

SMEs, including many startups, have an important economic and innovation role. Research shows that SMEs face many barriers in applying AI. The AiNed MIT AI call to bridge the gap between the knowledge base and application of AI innovations is supported from the AiNed Programme. This scheme will be repeated annually through 2026. Our website will keep you informed.

Published: 05/06/2023
Focus areas:
Sharing
Published: 05/06/2023
Sharing: