Should algorithms be regulated?
Labour MP says it may be time for government to get involved in Silicon Valley's secret sauce
Is it time to regulate algorithms? One Labour MP believes so -- and she's not the only one raising concerns about algorithmic accountability.
Chi Onwurah, shadow industry minister for Labour, told the Guardian that algorithms need to be regulated the same way their results are.
"Algorithms aren't above the law," she told the newspaper. She added: "The outcomes of algorithms are regulated the companies which use them have to meet employment law and competition law."
Onwurah admitted it's not easy to regulate algorithms when they lack transparency, saying the issue will be examined in a paper from Labour due in the new year.
The potential challenge to Silicon Valley's smoke and mirrors follows pressure on Google and Facebook to address the flood of "fake news" propagated by both services, and as more people find work in the so-called gig economy, with their employment influenced by algorithms created by the likes of Uber.
Such problems require algorithmic accountability, academics have argued. The question is how to give accountability to something that's by its nature confusing and more often than not the "secret sauce" behind a company's success.
"Algorithms are an area that in principle should be regulated, since they are ways of categorising people, things and actions and counting up the results," said Nick Couldry, Professor of Media, Communications and Social Theory LSE. "It matters a lot what is counted, and how the results are processed."
But he added that the need for regulation depends on the function. "I may care a lot less about algorithms that change what buying or listening recommendations I receive than those affect what appears to me as news, or affect how I am evaluated by others," he told IT Pro.
"Organisations should always take responsibility for how they represent the world and how they represent human beings, and this is no different," he added. "It is not so much a question of transparency (it may be hard for lay people to understand how an algorithm works even if they are told) as accountability. And underlying that, the question of whether the information on which the algorithm works should be collected in the first place."
In other words, algorithmic accountability may, in some ways, come back to data protection.Dr Alison Powell, assistant professor and director of Media and Communication at the London School for Economics, notes that theGeneral Data Protection Regulation stipulates that the function of an algorithm must be "made understandable to those they influence."
But she said that functional transparency may not be enough, and it may be necessary for the training data and other elements to be available for examination. "It is possible to regulate algorithms ex post - that is, in terms of their outcomes - but this is difficult in contexts where machine learning or other artificial intelligence techniques are used as there may be a gap between the claims made at the beginning of a process and the outcomes at the end," she explained. "Computer science scholars have proposed a strategy of 'procedural regularity' to address this. This process tries to make the steps of the algorithmic processing more regular so that there are ways of auditing and ensuring that the process isn't discriminating against people unfairly."
Top 5 challenges of migrating applications to the cloud
Explore how VMware Cloud on AWS helps to address common cloud migration challengesDownload now
3 reasons why now is the time to rethink your network
Changing requirements call for new solutionsDownload now
All-flash buyer’s guide
Tips for evaluating Solid-State ArraysDownload now
Enabling enterprise machine and deep learning with intelligent storage
The power of AI can only be realised through efficient and performant delivery of dataDownload now