Alphabet's DeepMind Should Clarify Profit Model, Panel Says

(Bloomberg) -- DeepMind, the artificial intelligence company owned by Alphabet Inc., should explain how it intends to make money from its work with the U.K.’s National Health Service, a review panel appointed by the company concluded.

The panel also said DeepMind’s health-care division needs to be more transparent about its relationship with the rest of the AI company, with Alphabet and with DeepMind sister company Google.

The public are likely to suspect that there must be an undisclosed profit motive or a hidden agenda without clarity about DeepMind Health’s business model, the panel wrote. “We do not believe this to be the case, but would urge DeepMind Health to be transparent about their business model, and their ability to stick to that without being overridden by Alphabet."

DeepMind, which is best known for creating software that could beat the world’s best human players at the strategy game Go, has a separate division, DeepMind Health, which applies machine-learning to health-care. It has worked with several NHS hospitals to design an app that alerts doctors to patients at risk of developing acute kidney injury. It also has several projects looking at whether software can detect other serious medical issues from medical imaging data as effectively or better than doctors can.

Patient Records

But the company’s work with the NHS has been controversial. The top U.K. privacy watchdog said last year that one of the hospitals working with DeepMind on its alerting app illegally transferred millions of patient records to the company.

When DeepMind formally created a separate health-care division, it appointed the panel of outside experts to conduct annual reviews of its work.

"The issues of privacy in a digital age are if anything, of greater concern now, than they were a year ago and the public’s view of the tech giants has shifted substantially," Julian Huppert, chair of the review panel, said in a foreword to the report. "We have been clear from the outset that ‘good enough’ is not good enough for a company with such a close relationship to Google."

Huppert said in an interview that while Mustafa Suleyman, DeepMind’s co-founder and head of its applied division, had offered assurances about its independence from Alphabet and Google when it comes to data sharing, DeepMind needed to enshrine these commitments in some sort of binding and public form. He said this was the only way people could be certain DeepMind’s independence would continue even if Suleyman eventually leaves the company.

Contract Texts

DeepMind has published the text of its contracts with the NHS trusts, with only minor redactions, showing that the company has largely done its work for free.

In response to the report, DeepMind said that its health-care division is continuing to define its commercial approach.

“We’re developing our longer-term business model and roadmap, and look forward to sharing our ideas once they’re further ahead,” DeepMind said in a statement. “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.”

Huppert, in the interview, said he was pleased to see that DeepMind intended to charge something for its work with the NHS because he wants it to be sustainable. But he also cautioned that DeepMind needs to let the NHS know as soon as possible what that model will be so that it doesn’t get locked into a system which it ultimately cannot afford.

Interoperable

He also applauded DeepMind’s commitment to using technology that is interoperable with other companies’ products so that NHS hospitals could, if they wished, switch to a different vendor without causing major disruption.

The company has also said in the past that it does not share data with Alphabet or Google. But the panel said DeepMind needed to provide more clarity around this policy, since DeepMind does store and process data in Google’s data centers.

Overall, Huppert said he was pleased with DeepMind’s efforts. "We are trying to hold DeepMind to a very high standard and we think they are doing an excellent job overall," he said. He also said that the review panel had gotten DeepMind to commit to a 12-point set of principals concerning their work that the panel can use to hold the company to account in the future.

To contact the reporter on this story: Jeremy Kahn in London at jkahn21@bloomberg.net

To contact the editors responsible for this story: Giles Turner at gturner35@bloomberg.net, Nate Lanxon, Kim Robert McLaughlin

©2018 Bloomberg L.P.