Competition law could help solve data-slurping monopolies, peers told
Also: Viscount Ridley says it's better bots spy on him because they won't tell the Mail
Information monopolies are a "vexing" problem, but data protection laws alone can't fix them, a UK parliamentary committee has been told.
Members of the House of Lords committee investigating artificial intelligence yesterday quizzed experts on how personal data should be owned, managed and used.
The peers voiced concerns about the emergence of data monopolies, where the biggest companies are gaining a stranglehold on increasing amounts of people's data.
At the hearing, Labour peer Anthony Giddens pointed to reports that 126 million Facebook users had viewed Russia-linked posts on the site, as a way to demonstrate the extent of such firms' influence.
"How would you think your way through that, in what is clearly, in relation to traditional corporations, close to a system monopoly?" he asked the panel.
Information commissioner Elizabeth Denham responded: "I think it is a vexing problem, and I do agree that people are getting more and more concerned about information monopolies that offer many, many kinds of services and link and collect all kinds of data."
But she argued the focus should be on competition authorities – not the data regulators.
"Data protection doesn't necessarily give you a way forward if the companies are collecting and using information in accordance with the law.
"Data protection doesn't stop mergers and acquisitions where the purpose is to exploit more personal data... I think the competition authorities have a role to play as more and more data is put together."
Fellow panelist Sandra Wachter, a researcher at the Oxford Internet Institute, agreed that only competition law could directly address monopolies, but suggested data protection regs giving more rights to data portability might benefit SMEs.
"An individual can take data from one data controller to another – that could enhance competition in a healthy way... without that need to regulate," she said.
University of Westminster ethicist Mercedes Bunz implied that it was too late to tackle monopolies of data ownership in some fields.
"The competition is nearly gone" in natural-language processing, she said, as Google, Facebook, Microsoft and IBM have a strong hold over such databanks.
She argued that there is "no easy monopoly or ownership of data" in the same way for transport or health data.
The committee also asked panelists what they thought of a recent government-sponsored report into AI, which stopped short of calling for specific regulation.
This was OK'd by the assembled panels, and participants in a second session said it was probably too soon for a specific AI watchdog or regulations.
"It feels premature to have a regulator for AI," said Olivier Thereaux, head of technology for the Open Data Institute .
"What's more useful right now is recognising that AI will be used across many sectors, many of which have regulators. They need help to figure out the impact of AI... for the moment that would probably be more impactful."
Javier Ruiz Diaz, policy director at the Open Rights Group, said that, given a lot of regulators needed to be informed about AI, it would be "more useful" to have an independent body that would offer advice and help.
The inquiry continues later this month.
Although the witnesses were clearly cautious of the way giant corporates are exploiting the data they hold on users, not all the committee members seemed so bothered.
Matt Ridley told the group: "Indeed, thinking about it, I'm a lot more comfortable with a bot knowing about my internet habits through my usage than a person, because a person might tell the Daily Mail... The fact it is a machine is to some extent reassuring to me, because I don't care."
Perhaps trying to reassure observers that not everyone on the committee was so blasé about the subject they're investigating, committee chairman Lib Dem Lord Clement-Jones later added: "Some of us may be less relaxed than Viscount Ridley on these matters."
Sponsored: Beyond the Data Frontier