Safeguarding democracy from false propaganda

Steven Kull and JP Thomas, Tribune News Service

The Founding Fathers asserted that elected officials should listen to and be influenced by the views of the electorate. As James Madison said, “it is the reason, alone, of the public, that ought to control and regulate the government.” However, the means for government officials to hear from the people are limited. Elected officials receive emails, letters, phone calls and input at town halls, and some agencies occasionally ask for public comments on complex regulations. Americans express very low levels of confidence that they influence the policymaking process and that elected officials understand the views of the people, undermining trust in the democratic process.

AI can generate vast numbers of communications to policymakers masquerading as citizen input. When the Federal Communications Commission in 2016 took public comments on whether to retain net neutrality, the New York Attorney General later found nearly 18 of the 22 million comments were fabricated, using generated fake names, or real names without consent. A large portion was generated by the broadband industry, which wanted fewer regulations. Among the genuine inputs, 98.5% favoured retaining net neutrality.

Researchers conducting a recent study for Brookings sent 32,398 emails to legislative offices, some written by citizens, others generated by AI, which can deliver thousands of letters that sound genuine seconds. The study found legislative offices could not discern which were fake. We are just seeing the beginning of what will become a flood of false inputs drowning genuine input and further undermining public confidence.

So what can be done? It may be that the collapse of a flawed system requires not a superficial repair, but a major upgrade. In this country — and other democracies — universities, nonprofit organisations and governments have been using methods for consulting citizens on important policy issues. These public consultation efforts recruit representative samples that mirror the public as a whole, making it possible to discern consensus. And they verify that those who engage are, in fact, real citizens using modern technologies.

Further, they go beyond polls, which are limited to a narrow range of topics on which the public already has enough information to give meaningful input. For example, in “public consultation surveys” developed by the Program for Public Consultation at the University of Maryland, up to several thousand verified citizens go through a transparent online process in which they get a briefing on a policy proposal the government is debating and evaluate arguments for and against. All content is reviewed by policy experts on different sides of the debate. So when citizens make their policy recommendations, it is an accurate expression of their values and priorities. Voice of the People also conducts these online surveys in congressional districts and then brings together constituents who took the survey and representatives to have informed discussions on the issues.

Other programs take smaller groups of citizens through more in-depth processes — called “citizen assemblies,” “citizen juries” or “deliberative polls” — that can last several days. In addition to getting briefed, these “mini-publics,” which can range from several dozen to several hundred, meet and discuss issues with experts and each other before coming to conclusions. Asked how they feel about using public consultation, a very large majority of voters approve and say it would improve their confidence in government. They also say they are more likely to vote for candidates who commit to engaging their constituents in this way, with more than four in 10 saying they would even be ready to cross party lines.

What is striking is that when citizens engage in these public consultation processes, they are far more likely to find bipartisan common ground than Congress. From public consultations with nearly 100,000 Americans, Voice of the People has identified over 200 points of bipartisan agreement, in a wide range of policy areas, on which Congress has been gridlocked. While privately funded efforts are making a meaningful contribution, if policymakers truly want to listen, they should create a government-funded institute to consistently consult the people.

We stand at a crossroads. AI has opened up new possibilities for distorting the democratic process by generating fake voices in the service of special interests. At the same time, democratic, technological innovations have opened up the possibility of not only verifying the voices of representative samples of citizens but also giving them tools to deliver meaningful input to their elected representatives. Which use becomes the norm will set the future of democracy itself.