AI in Voting: Our Approach to Civic Tech

Civics Beyond the Ballot

Civics is all too often simplified to the action of voting, despite comprising the connective tissue of our society. From governing how our different social groups should interact with each other, to building and maintaining our roads and bridges, civics is the activity of civilization. But it’s so pervasive that it fades into the background of our consciousness.  

For many people, civics doesn’t feel present day to day. Only when politicians come back begging for more votes do we remember our “one” civic duty: voting. And if we’re lucky, some of us will actually vote. 

At the same time, a new technology is permeating through our lives: artificial intelligence. And because civics touches everything, AI is inevitably becoming part of how we engage with democracy - whether we’re ready for it or not. But we do like one aspect: the promise of what AI can bring to voting.

With the right use of AI, you can see past the narratives that any one candidate or party is trying to push, you can access and compare all of the information about candidates in one easy place, and you can be more confident in who and what you are voting for. Critically, by improving voting, especially through transparency and reduced barriers of entry, the right usage of AI can free us to consider the other aspects of civics in our lives. 

The Promise and Peril of AI in Democracy

The danger with AI in voting is that the “right usage” is fairly narrow, and bad actors are salivating at the opportunity it presents to change and even limit how you vote. The current examples in politics with deep fakes and targeted ads are not even the worst – subtle manipulation of the information you consume can be just enough to change your mind and tip the scales in ways you wouldn’t even realize after the fact. That is why responsible, helpful, and transparent AI in voting is critical in staving off the inevitable horror of unchecked AI involvement with voting. This is one of the driving forces for VoteMate, your nonpartisan voting  assistant that provides clarity – not narratives.  

VoteMate’s Five Guiding Principles

Along with displacing harmful uses, we’re committed to making sure VoteMate always stays responsible and genuinely helpful.

Our first approach is transparency, both in how we bring AI use cases to civics, and in how our users interact and interpret our AI systems. This post is the start of how we are sharing our approach, and our online material will continue to detail our position and efforts as an organization.

As a platform, we are also committed to ease of use and understanding of what AI is doing for you. We offer a chat experience to understand your priorities, but in our ballot builder, we do not have any algorithms, there are no calculations beyond simple addition, and there is no unexplainable matchmaking. We offer simple comparisons on stances that are easily interpretable, explainable, and verifiable to make sure you agree the candidate or measure aligns to your criteria. 

Our second approach is accountability. We have an idea for how to use AI responsibly in voting, but we need your help to let us know what is not working, or what can be done better. We plan to have a general ticket submission process, as well as feedback submission for each message from VoteMate to capture the good and the bad so that real humans can make sure our AI serves you in the way you want. 

Our third effort is security and privacy. We will never sell your personal information, and we are using industry standard security practices, reviewed by world-class security experts, to ensure your data only belongs to you. You can delete your data held by VoteMate, and you can even interact with the platform without signing in to ensure your data is not saved in the first place. 

Our fourth task is reliability and safety. We’ve partnered with other civic organizations to bring you the most accurate data about elections and ballots so you can be confident in your voting habits. But VoteMate offers much more than what you can find on your state election body’s website.

We are developing a robust evaluation and monitoring system to ensure VoteMate provides accurate and complete information and a process that can be trusted by every voter. And when something might go wrong, we will be positioned to respond quickly to take responsibility and remedy the situation. 

Our last tactic is to be fair and inclusive. In order to prevent nefarious AI manipulation, VoteMate needs to be everywhere and able to serve everyone, no matter their background or views. This means no political bias, an accessible platform, and we hope to support multiple languages in the future so that VoteMate can meet each voter where they are. 

Building Trust in Civic Tech, Step by Step

Taken together, these five approaches stand as our principles for how we are introducing AI to voting. They mirror the principles many top companies and organizations adopt, and we are lucky to work with industry leaders who have already proven their guiding strength.  

With these principles, we believe we can fulfill the promise of voting with the power of AI and avoid the potential horror of unchecked AI influence on the voting process. We don’t believe we can un-skew any of humanity’s biases with VoteMate, but we can prevent AI from skewing us further. And if we can successfully carve out enough space with AI-enhanced voting, we may be able to notice our other civic duties and spend more quality time on them.