Three Paths Towards Global Governance of Artificial Intelligence

This month, Amazon.com Inc. shut down a project it had been developing for four years — a recruitment tool driven by machine learning. The concept was a simple and appealing one: at its core, the project aimed to develop an algorithm that would sort incoming job applications to isolate the short list for managers to use in making their final selections.

Anyone who has been involved in such a process knows that isolating the top five or ten resumés from dozens of applicants is a time-consuming job. Any process that brings logic and speed to this stage of the recruitment chain can only be a good thing.

But algorithms are only as good as the data used to drive them, and in the Amazon case the machine “learning” was based on patterns in applications submitted to the firm in the previous ten years. Since Amazon, like all tech firms, is male-dominated, the algorithm taught itself that success equated with male-oriented activities and language, so phrases like “women’s chess club captain” or the names of all-women colleges, were penalised by the algorithm in ways that technical fixes could not remedy.

This episode exemplifies many of the appealing and unappealing features of the data-driven era that is rapidly coming upon us. The appeal is intuitive. If tasks, even complex ones, can be completed rapidly, accurately, and effectively by a tool that improves itself over time, then a range of tedious and expensive chores currently handled by humans can be outsourced, not to cheaper humans far away, but to software embedded right in your office.

However, there are also three broad sets of issues that give pause for thought and where globally agreed norms and rules — in a word, governance — is called for.

First, algorithms, whether static or of the machine-learning sort, are not value-free. The data underlying them and the formulae that make them function, think, and transform over time embody the biases of history and that of their designers.

This means that algorithms should be subordinated to the same kind of universal ethics regime that governs human and state behavior: something similar to the Universal Declaration of Human Rights (UHDR). Though the UDHR is often violated, it creates an aspirational global standard. It acts as a guide to draft national and sub-national legislation, as a framework to assess (and sometimes “name and shame”) its application, and ultimately, as with the International Court of Justice, to penalise its violation.

The recent G7 statement on artificial intelligence (AI) serves as a good starting point for a more global discussion on the ethos that we want driving transformative technologies in general and AI in particular.

Earlier in the summer, Amazon was in the news for another feature of footloose multinational firms — the taxes they pay. The high-tech economy, of which AI is a part, is driven by proprietary technology, most of which is created in a few pockets around the world. It is the nature of the innovation economy to privilege first movers, strategic behavior and economies of agglomeration.

As a result, despite many unknowns about what the future holds, one thing is clear: absent of significant public policy responses, income and wealth disparities will worsen, between countries and within them.

Since so much intellectual property is generated by multinational firms, the profits from technological advances will (correctly) accrue to them.