In a recent review, an independent group of panelists raised questions about UK-based AI company DeepMind’s business model, relationship with parent company Alphabet, and potential to monopolize its industry.
The Google subsidiary has been partnering with the National Health Service for the last few years and working with the system to implement its AI technology. While the company has been growing both in name recognition and partnerships, questions arose following that led the panel to pinpoint specific areas for the company to examine.
The recent report was the end result of a yearly review by an independent panel DeepMind created in 2016. The panel, chaired by Dr. Julian Huppert, director of the intellectual forum at Jesus College Cambridge, is made up of representative from multiple related industries.
One of the areas that the report zeroed in on was the startup’s relationship with Google and parent company Alphabet. The group posed questions about data sharing and security between the companies.
Currently DeepMind states that data acquired through their site will never be “connected to Google accounts or services.” But panelists called on the company to spell out a more in-depth explanation as to what exactly that term means for the health subsidiary.
“We recommend that DeepMind Health should be transparent about its business model,” the panelists wrote, “that DeepMind Health specify exactly how they will work with other elements of Alphabet, and what data could ever be transferred to them.”
The panel noted that DeepMind’s revenue source is still murky, another cause for concern.
“It is important for the public to have reassurance about DeepMind Health's business model,” the reviewers wrote. “They want to know that DeepMind Health’s revenue is — or will be — coming from a source they consider appropriate. They also want to know that any promises made now by DeepMind Health will be kept in the future and that if this generation enjoys a service for little or no cost, that the next generation will not pay the price in excessive costs or lack of control further down the line.”
Selling even raw depersonalized data or using them for advertising purposes would create a “deep concern” for the panelists. While the company doesn’t seem to be interested in this type of model, the panel questioned exactly how the company plans on making money.
“We have had detailed conversations about DeepMind Health’s evolving thoughts in this area, and are aware that some of these questions have not yet been finalised," they wrote. "However, we would urge DeepMind Health to set out publicly what they are proposing. If not, there is a real risk that the public will assume that the work is driven by a purely profit-making motivation and will have stronger grounds for cynicism about how data and machine learning will be used.”
In the company’s response to the report it pledged to examine business model-specific questions.
“We’re developing our longer-term business model and roadmap, and look forward to sharing our ideas once they’re further ahead. Rather than charging for the early stages of our work, our first priority has been to prove that our technologies can help improve patient care and reduce costs,” DeepMind wrote. “We believe that our business model should flow from the positive impact we create, and will continue to explore outcomes-based elements so that costs are at least in part related to the benefits we deliver.”
Regardless of the questions around the revenue source, DeepMind is gaining steam in the UK. But that could create new problems, according to panelists.
“A separate issue relates to the extent to which DeepMind Health, even apart from its connections with Alphabet, could find itself in a position of being able to exert excessive monopoly power,” the panelists wrote. “There are many examples in the IT arena where companies lock their customers into systems that are difficult to change or replace.”
There is a fear that DeepMind could make it difficult for providers to switch to other IT services. While a monopoly like this wouldn’t be good for the public interest, panelists said, transparency and a commitment to interoperability could prevent this outcome.
Other issues raised by the panel involved legal, clinical, and security concerns. A new review will be released next year.
“By holding us to account, recognising where we’re getting it right and challenging us where we can improve, they’ll help us to do a better job for patients, nurses, doctors, carers, families, and all those who rely on healthcare systems around the world,” DeepMind wrote in their response to the review.