Technical / Methods Track
How to Achieve Both Transparency and Accuracy in Predictive Decision Making: an Introduction to Strategic Prediction
“When an algorithmic system holds sway over a person’s life, the person has an incentive to respond to the algorithm strategically in order to obtain a more desirable decision.” This is the mantra of the rapidly growing research area of strategic prediction, which holds the promise of making transparency more palatable to decision-makers. Algorithmic transparency is sometimes thought of as the holy grail of just machine learning, but allowing someone to perfectly understand an automated system makes the system prone to “gaming”. This can in turn threaten the accuracy of the system, which is the decision-makers’ utmost priority. To this end, the field of strategic prediction aims to answer a fundamental question: “how can we create machine learning systems that are accurate even when made transparent, and thus gamed by individuals?” And even opaque systems are rarely perfectly opaque, so strategic prediction is relevant in almost all predictive decision-making scenarios. Our tutorial provides an overview of the ideas and the toolkit that computer scientists have begun building to allow decision-makers to simultaneously provide accuracy and transparency. The dissemination of these ideas into the communities of policy-makers, lawyers, and other stakeholders is crucial—otherwise, civil liberties advocates and the institutions they seek to reform will be stuck in a fruitless stalemate between “transparency” and “accuracy”. Our tutorial builds in these stakeholders an ability to assess both the tension between these objectives and an intuition for the numerous approaches for satisfying them together. Our aim is that our audience gains the grounding to have an informed discussion about strategic prediction in a nontechnical context so that they can introduce these ideas into their own advocacy efforts and launch complementary explorations in their own academic literatures.