SAN FRANCISCO — A new partnership between a United Nations agency and a data mining company points directly to the tradeoffs humanitarian organizations must weigh when entering partnerships with the technology industry.
Enrica Porcari, director of technology at the U.N. World Food Programme, called the new partnership with data mining company Palantir an example of how the agency is “transforming humanitarian work one partnership at a time.”
“We need a public discourse around where to draw the line in very concrete cases between misuse and missed use.”— Robert Kirkpatrick, director, U.N. Global Pulse
WFP seeks to make the vast amount of data it generates accessible across the organization in order to streamline the delivery of food and cash, but data privacy activists say the partnership with the Silicon Valley-based company puts the recipients of WFP aid at risk.
Privacy International recently published a report outlining how humanitarian organizations can apply the “do no harm” principle to their work in the digital era. As organizations explore how new technologies might apply to their work, they have to consider the potential for unintended consequences, as well as the consequences of not partnering at all.
Robert Kirkpatrick, who leads U.N. Global Pulse, an innovation initiative of the secretary-general, said a major challenge around the use of big data is that risk of harm related to privacy needs to be balanced against the risk of other harms that could be prevented by using the data. He said current regulations often focus solely on privacy risk and don't leave adequate space for beneficial uses of big data for the public good.
Some have argued that a good starting point would be to educate the public about the relationship between privacy and utility of the data we produce in daily life, he added. The topic was raised at the World Economic Forum Annual Meeting in Davos, Switzerland in late January, where representatives of UN Secretary-General António Guterres' High-level Panel on Digital Cooperation held a roundtable discussion.
“We need a public discourse around where to draw the line in very concrete cases between misuse and missed use,” Kirkpatrick said. The statement was met with nods of agreement from leaders around the table.
Next month, Kirkpatrick will attend the Mobile World Congress in Barcelona, Spain, a conference hosted by the GSM Association, a trade organization representing the interests of mobile network operators. There, policymakers, international organizations, and industry leaders will discuss ways to work together to use mobile data for global goals while also protecting the privacy and security of subscribers.
“We’re looking at food companies, pharma, retail, all these different industries and saying: ‘How do we get them to the position the mobile industry is in?’” Kirkpatrick told Devex.
“We have bilateral partnerships, but how do we scale up to an industry-wide strategy, based on the success stories and case studies that already exist and the obvious opportunities to contribute to the global good?”
There is a need to open up the kind of real-time information available across industries in a way that mitigates the risks — but also uses data to mitigate other risks, he said.
“Data can help to shrink the size of the problem,” Jonathan Stambolis, founder and CEO of the big data and artificial intelligence company Zenysis Technologies, said at Devex’s Prescription for Progress event in San Francisco, California.
He pointed to the work Zenysis engaged in with the Ministry of Health in Ethiopia to optimize the country’s nationwide measles vaccination campaign. They combined data sources that previously operated in silos to understand in which districts it made sense to vaccinate children under 15 versus just those children under 5.
“That data-driven, targeted, ‘shrink the size of the problem’ approach saved an estimated $30 million, and that’s money that could be put into other lifesaving programs in the health system,” Stambolis said.
Stambolis led international partnerships for Palantir before starting Zenysis, which aims to support governments in their work to address global health and international development priorities.
Palantir is linked with the U.S. Central Intelligence Agency, which invested in the company through its venture fund and is a customer along with other U.S. government agencies. Palantir is credited with supporting the U.S. government in its efforts to track down Osama bin Laden. A fake vaccination campaign sent health workers to collect DNA in the neighborhood in Pakistan where the Al Qaeda founder was hiding.
While the operation succeeded, it eroded trust against vaccination programs, which public health officials say could set their progress back by decades.
“While the misuse of humanitarian or public health data may be rare, the consequences can be severe,” Stambolis told Devex via email in response to a question about data risk.
Discussions on privacy, bias, and security need to happen more globally, and at the WEF meetings in Davos, leaders across sectors can build partnerships to mitigate these risks.
The conversation on ethics and technology is underway in Silicon Valley and continued in Davos, but it is one that needs to happen more globally, Kirkpatrick said, emphasizing the importance of discussing risks as well as opportunity costs.
"While it is imperative that global development organizations partner with the private sector to unleash the power of their data to save lives, it is also imperative that they elevate privacy and information security to first-order priorities and partner with organizations whose values and business models are aligned with their humanitarian mission,” Stambolis said.
Update, Feb. 7: This story was updated to clarify remarks made by Robert Kirkpatrick, add additional context, and clarify that the roundtable discussion was held in support of U.N. Secretary-General António Guterres’ high-level panel on digital cooperation. The story was also updated to clarify that Jonathan Stambolis responded to Devex via email regarding a question on data risk.