An Interview with Darrell West: Policymaking in the Era of Artificial Intelligence

Sep 10, 2020 by

Turning Point: Policymaking in the Era of Artificial Intelligence by [Darrell M. West, John R. Allen]

Michael F. Shaughnessy –

1) Darrell, you have just co-authored a book entitled” Turning Point: Policymaking in the Era of Artificial Intelligence”. First, let’s get on the same page- how do you define “artificial intelligence”? 

AI is automated software that learns from data, text, or images and makes intelligent decisions based on that analysis. Its key features are intelligence, learning, and adaptability. AI already is being deployed and making decisions in healthcare, education, transportation, e-commerce, and national defense, among others. Students are being assigned to schools via AI algorithms. Healthcare is being assessed through AI-driven data analytics. On e-commerce sites, AI can monitor your purchases and ad viewing, and offer you product recommendations before you even know you want a certain item. AI is the software in autonomous vehicles that integrates camera information and keeps the car in the correct lane. And in national defense, there are autonomous weapons systems that can act independently of human guidance.

2) Why do you feel we are at a “turning point”?

We are at a turning point between utopia and dystopia in that the technology has advanced rapidly in recent years and it can usher in a positive scenario of good health and widespread prosperity or a dystopian world of unfettered surveillance, income inequality, and pervasive bias. The crucial variable in determining the future is public policy. We have the means to bend the curve in a positive direction and address major problems or seeing things go quickly off the rails if we fail to take meaningful action.

3) Now, when you talk policy- are you talking educational, medical, social or just over-all policy making”?

By policy, we mean authoritative actions of government designed to make sure AI decides things in accordance with basic human values. We need AI that is fair, unbiased, inclusive, accountable, transparent, and safe. Public policy can ensure that AI develops in accordance with these values or if no action is taken, that it increases inequality, furthers bias, invades privacy, and widens social, political, and economic divisions within society. We control the future of AI by the choices we make in coming years.

4) Could “artificial intelligence” have predicted COVID-19 and the ramifications and repercussions of it? And perhaps avoided it? Could “artificial intelligence” help with treatment decisions regarding COVID-19?

AI could not have predicted a COVID pandemic but in conjunction with data analytics, it has provided a valuable service in tracking the disease spread, enumerating racial and geographical disparities, and modelling the impact of particular remedies. Right now, it is being used to help scientists find new chemical compounds that can form the basis of new drugs or vaccines. It can scan the scientific literature more quickly and efficiently than humans and suggest treatments that could prove helpful.

5) I am somewhat skeptical of Artificial Intelligence as sometimes, as we say “life happens”. Witness the recent riots and fires and damage and destruction to various cities in the U.S. How can “Artificial Intelligence” help both law enforcement and senators and congressmen/ women to address these issues?

We shouldn’t expect AI to resolve deeply-rooted value decisions in society and politics. That is well beyond current technological capabilities. Humans have to make decisions on which values it wants to maximize and then algorithms can make decisions that reflect those values. It can’t happen the other way around and we don’t want AI deciding which values are most important to humans. That would not work out well for humanity.

6) In your mind, what ARE the benefits of artificial intelligence? And are they being used?

AI can take over boring, repetitive, or dangerous tasks from humans. It can relieve humans of routine decisions that aren’t contentious and don’t require high level value decisions. In so doing, it can improve human efficiency and performance, and augment basic human skills. But AI has to be pointed in the right direction. We have to instruct our algorithms what goals we wish to further and then make sure the AI acts in accordance with our values.

7) What have I neglected to ask? You present a policy blueprint. What are the basic things that need to happen in order to move forward with AI?

We have a number of ideas on how to develop responsible AI. It involves developing anti-bias rules, improving digital access, restoring the U.S. Office of Technology Assessment so government leaders have access to expert technology advice, developing a national innovation plan, and requiring AI impact statements for large-scale, publicly financed AI projects. These actions will help us anticipate possible problems and figure out how to mitigate them before they become big problems.

Print Friendly, PDF & Email

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.