The challenges of regulating artificial intelligence
Robots and regulation
27 April 2017
Did you know that cars can learn how to park themselves? Or that an MIT algorithm can now predict the future… or at least 1.5 seconds of it? These are amongst the latest innovations that were unveiled in Barcelona in December, at the 30th Neural Information Processing Systems conference, or NIPS 2016, the leading conference for machine learning and artificial intelligence.
Artificial intelligence is progressing at a rapid pace, infiltrating an increasing number of industries and sectors and posing significant challenges for both companies and the bodies that regulate them.
As a law firm, Clifford Chance is no stranger to artificial intelligence. We, like many businesses, are introducing intelligent technology into our own organisation to drive improved quality and efficiency. However, in addition to our own adoption, it's also our role to advise our clients on the issues that surround the implementation of new technology in their own sectors.
In recent years we've seen technology disrupt every business, rendering the very definition of a 'technology business' a slightly tortured one. As this has happened, it has driven a wave of collaborative arrangements between traditional firms and technology start-ups, and given rise to important commercial and regulatory considerations. For instance, firms and technology companies need to determine the appropriate model for their collaboration exercise – is it a joint venture, an acquisition, licensing or something else? Related to this, the allocation of assets (such as IP and customers) and liabilities between the technology provider and firm will need to be balanced. Finally, all of this must take place within the regulatory schema, granting the appropriate access to the regulator and protections for consumers.
As technology is applied to regulated industries, businesses on both sides of the disruption can struggle to understand their obligations. This is perhaps unsurprising when the regulation itself was often written before these new technologies were even invented.
AI and the law
With regard to artificial intelligence, advances in the field mean that the predictions and decisions of algorithms are already entering into a variety of industries that operate under a comprehensive regulatory framework, posing a number of significant legal questions.
Legal and regulatory frameworks typically operate around a clear sense of who is acting, what their mindset was at the time of the action and where the act takes place. These are all fine principles in theory but as soon as this approach is applied in the context of AI, it is clear that new ways of thinking are required. For example, who is liable for accidents if cars are driverless? What recourse might an individual have if they are refused insurance based on computer analysis of their social media accounts? Who should be liable for trading losses in a flash crash triggered by a US bank using a news-reading algorithm licensed from an Indian software company? Most fundamentally, how does existing regulation apply when agency over regulated activity is taken away from people and put into the hands of machines?
These are just some of the questions that have to be considered afresh and the issues that we continue to confront as we advise our clients on how best to apply the existing legal and regulatory framework to a rapidly changing world.
Creating technology-cognisant regulation
Navigating existing regulation however is just one piece of the jigsaw, and arguably a much smaller task than that facing the regulators, who must formulate a new architecture to realise the potential of new technologies whilst protecting consumers and maintaining effective markets.
Whilst regulators must be aware of the risks presented by new technologies, the opportunity to adopt more efficient ways to regulate is a prize worthy of pursuit by the regulators and the regulated alike. As advisors to both, it is here that we as lawyers have a particularly important role, guiding that necessarily collaborative process.
In the interest of such collaboration, NIPS 2016 included the conference's first ever symposium and award, sponsored by Clifford Chance, dedicated to machine learning and the law, seeking to bring together the worlds of technology and law to inform better policy for the future.
There is no doubt that artificial intelligence is capable of transforming nearly all industries, creating efficiencies that, handled in the right way, will improve the goods and services available to consumers and businesses. To achieve these benefits it is imperative that the worlds of technology, industry, regulation and academia continue to collaborate over the appropriate legal and regulatory frameworks that make them possible. It is here that we can all contribute to achieving a bright future.
This article was originally published in the BritishAmerican Business Network magazine. To view the original article, click here.