Sharing is caring: Surviving the AI economy through Data Trusts

A novel legal construct is seeking to balance the dual needs of fast-paced innovation and privacy

AI and analytics has a “seesaw” problem. Innovation in AI requires access to rich and expansive datasets, data which is being created all of the time by a raft of entities. But personal and corporate data, if not handled correctly, risks exposing individuals or organisations. If we limit the amount of data entering the innovation funnel, the seesaw pivots in the opposite direction, hamstringing innovation.

A new legal construct is seeking to balance the dual needs of fast-paced innovation and privacy by allowing data owners to cross-pollinate data securely and ethically. It’s called a Data Trust, and interest in them is rising where data sharing is important, such as in smart cities, health data ecosystems or collaborations between business and academia. 

“A Data Trust is essentially a legal construct that enables various data providers to share their data in a secure, compliant and ethical way,” explains George Zarkadakis, digital lead at Willis Towers Watson, where he is leading the development of a Data Trust to connect a number of the global advisory firm’s clients.

Rising tides

The business demand for Data Trusts, a term first coined by the UK Open Data Institute, can be traced to AI’s centrality to the 21st-century economy. This economy is built on data, but the foundations are rocky. Good quality and varied data is scarce, and the data landscape is fragmented. Where data is up to scratch, a range of regulatory, technological and ethical issues keep it siloed, shielded and unable to blossom. 

“To survive and thrive in the AI economy, companies must make considerable investments in how they collect, store and handle data in a collaborative way,” says Zarkadakis, who offers the example of an airline, a hotel chain and an insurer:

“They all have data about their customers, but what they lack is a way to understand the whole customer journey, from the moment people decide to travel, to booking their flight and accommodation, to purchasing an insurance policy. The best solution would be for those three businesses to share their data in order to serve their customers better and increase their competitiveness.”

Tag team

Unless there is a mutually-incentivised framework in place, there’s no chance you’ll catch any of the three organisations swapping their prized data. 

The first challenge is ensuring data anonymisation and security. Although a number of technologies provide these functionalities, there’s still a lot of work to be done to establish a standardised technology stack. Then there are policies and data governance frameworks (not to mention processes for auditing and reporting the use of data)  that have to be considered. At this early stage what constitutes a compliant and secure ecosystem will differ from case to case. 

Join the data

Aside from security and compliance, a necessary characteristic of any data trust is the presence of robust incentives. As mentioned above, in the context of business, there is somewhat of an existential incentive as AI becomes the nucleus of the economy. Data Trusts breathe life into AI agendas by improving “data liquidity”. In other words, how quickly, efficiently and effectively data is sourced, used and deployed to solve business problems. 

One of the biggest concerns raised by members of Willis Towers Watson’s pilot was the risk to competitiveness. Fierce collaboration is perhaps the most counterintuitive quality of Data Trusts. Since when do competitors share their prized jewels? While a legitimate concern in some circumstances, the criticism is a little off the mark. Data Trusts aren’t vying to subsume all organisations into one data web, but to connect those where the connection is needed.

“We worked with our partners on specific use cases where the dividends of collaboration were not only significant but also unrealisable unless there was a Data Trust,” explains Zarkadakis. 

At Big Data & AI World London, Zarkadakis will share his learnings from the Data Trust pilot he has helped to develop, and detail how issues of commercialisation, privacy and security are being addressed. He will also sketch his “manifesto” for Data Sharing — his own guidebook for navigating this new collaborative world.

“The Data Trust is a way to bring huge commercial returns to our data partners while delivering significant and positive value to society at large. The future data economy will be a mostly collaborative economy, and this can happen only when there is a win-win for all stakeholders. It would be this win-win, for business, public institutions, cities, countries, and citizens that will inspire the manifesto of our Data Trust.”