Keystone Strategy

The Ethical Algorithm: A Q&A With Co-Authors and Keystone Experts, Aaron Roth and Michael Kearns

In their new book, The Ethical Algorithm, Aaron Roth and Michael Kearns explore the emerging science behind creating ethical algorithms and the mathematically precise ways of analyzing algorithmic behavior.
November 1, 2019   /   6 Minute Read
The Ethical Algorithm, Aaron Roth and Michael Kearns

Over the past decade, algorithms have become the backbone for providing users with advanced services and new functionality. As algorithms make our lives more efficient, more entertaining, and even better informed, they also run the risk of violating our basic individual rights with how they collect and use our personal data to make decisions. In their new book, The Ethical Algorithm, Aaron Roth and Michael Kearns explore the emerging science behind creating ethical algorithms and the mathematically precise ways of analyzing algorithmic behavior. In this Q&A, Keystone’s Gautam Basak (GB) checks in with the authors on the eve of their book release (Nov 1, 2019).

Gautam Basak (GB): What inspired you to write about two seemingly unrelated topics – Ethical Values and Algorithms?

AR + MK: The primary point of the book is to show people that these topics are not unrelated. When human beings are involved in important decision-making pipelines – education, medical treatment, lending, and criminal justice — we expect that these decisions are made ethically. Over the last decade, we’ve started to move algorithms into parts of these decision-making pipelines and that is going to have ethical consequences, whether they are made explicit or not. Our book and research argue that if we want to embed these nuanced social norms as constraints on the behavior of algorithms, we need to think about them in a mathematically precise way. The Ethical Algorithm focuses on the emerging science of how to do that.

GB: Can you describe the rise and importance of algorithms today and implications going forward?

AR: There are several key takeaways: Consider the methodology we propose to think carefully and precisely about definitions, and what we really want from our systems — and then trying to design algorithms that satisfy these notions, and soberly exploring the tradeoffs involved. It is hard, but can be done. There’s an emerging community of scientists and engineers doing exactly this.

Scientists and engineers will need to build the necessary tools, regulators must understand the science in order to legislate sensibly, and business leaders will need to prioritize their goals. Stakeholders in every industry will need to understand the costs and benefits of new technologies in order to engage in informed advocacy and decision making.

MK: We deliberately set out to write a book for the widest possible audience. Although everything we discuss has rigorous underlying science, there’s not a single equation in the book yet we have managed to bring the science alive by giving varied, real-world examples.

GB: How was it teaming with each other on co-authoring this book?

AR: It was great! When folks ask me how long it took to write the book, I always tell them that it went twice as fast as it otherwise would have, because there were two of us! I’d do it again for sure.

MK: Yes, in my experience, the time to jointly write increases with the number of authors, not decreases, but this was entirely the opposite and I wasn’t surprised at all — Aaron and I are simpatico in so many dimensions.

GB: How can companies partner with you in driving ethical use of software and data in their businesses?

AR: We both think of technology transfer as part of our jobs, on top of basic science, and both spend time consulting about practical applications of the science we talk about in the book. In the past I’ve worked with Apple during their rollout of differential privacy in iOS10 and with Facebook on their Election Research Commission, helping to facilitate giving social scientists the data that they need to study e.g. the spread of fake news without violating the privacy of Facebook users. We’re interested in opportunities to put this new science into practice.

MK: In addition to tech industry consulting, I have a long “shadow” career in quantitative and algorithmic trading on Wall Street, where the concerns we write about are rapidly being raised by regulators asking financial institutions about the ways they are using machine learning and AI, and whether those methods are beneficial to the institutions themselves or to society.

If you are interested in learning more about ethical algorithms and the science behind measuring algorithm behavior, order your copy of The Ethical Algorithm here.

Find out why top tech firms, Fortune 500 companies, and global law firms partner with Keystone.