Digital Democracy in a time of Misinformation, Automation and Machine Intelligence

Digital Democracy in a time of Misinformation, Automation and Machine Intelligence

Innovation and technology have always been drivers of electoral success. Franklin D Roosevelt made use of radio in 1932 in election campaigns and political contests. Remington Rand’s UNIVAC computer was debuted to help predict the election results in 1952.

In 1984, President Reagan’s re-election committee developed a computer system to test its campaign strategy on a daily basis by analyzing polls, census data, economic conditions and the assessments of its top political operators. In 2008 Barack Obama had a presence on 16 different social media platforms which he used to great. Intelligent campaigns will always dive into tech advancements to get ahead.

In 2014 while researching and studying data science, I decided to apply the knowledge I’d garnered so far for the approaching 2015 General Elections in Nigeria. Putting a team together to work with me, by June 2015, the project had published hundreds of visualizations and analysis with a people reach of almost a million online and 140,000 weekly offline (newspaper) reach.

No alt text provided for this image

The trick was to find questions asked by citizens or misinformation peddled by malicious entities leveraging an open source social analytics tool called Keyhole.co. Whatever questions we saw that had the highest counts or misinformation that turned up, we answered or corrected using credible information from the Nigeria’s official Electoral body the Independent Electoral Commission (INEC) resources and other credible civil society organizations, then converted to infographics and shared.

Citizens wanted to know more about campaign finance and spending, agendas and manifesto, ballot qualification, voter registration, voting procedures, Permanent Voter’s Card (PVC) distribution, electoral violence, ballot box snatching, polling stations, Electronic Card Reader (ECR) testing, components, advantages and opinions surrounding its utility. And we leveraged a visual and easy-to-communicate medium to provide them access knowing fully well that a well educated citizenry with the factual information is good for democracy.

No alt text provided for this image

In 2018, in the build up of the 2019 elections while thinking of the best use of automation and algorithms for elections, we had started collecting, analyzing and structuring the data from the 2015 elections. With a team of developers working with the AI Research company I co-founded called AI Envoy and Robotics, we began building a bot that will foster access to data in a much more conversational way. Finding a specific data or information was hard, we wanted to build a use case for Conversational AI for elections to automate information discovery. So tinkering with Google’s Tensor Flow, Robot Vee was born.

And it did everything we wanted it to do. Answered questions correctly, identified Nigerian state names no matter how badly pronounced. Provided answers to queries in less than 5 seconds. We thought we were done.

No alt text provided for this image

After several months of training of the bot to recognize more terms and testing it for device suitability, we realized that only a few phone models and versions could actually use it. We had an option of deploying it from Tensor flow to a more friendly messaging platform like Whatsapp, but that came with its challenges, we will not be able to push notifications as per Google’s Push Policy. So we resorted to start afresh, building this time with Chatfuel. The result was impressive. Read more about its features here.

No alt text provided for this image

We changed the name from Robot Vee to Voteron (pronounced Vote:ron) which we had always wanted to use but couldn’t since Google also had issues with one word naming. We coined the name from the 90s cartoon called ‘VOLTRON’ — the defender of the universe because we wanted people to see the bot as something to help them defend their votes, facts, and factual information in a time when fake news and information is rife. This time we succeeded, were able to include Live Updates which we couldn’t add on Tensor Flow. You can find the bot here.

Building this technology was an eye opener for both the team at AI Envoy and Orodata. We have been fighting for misinformation since 2014. We have learnt a lot, we understand that responsible reporting and factual information is crucial for people to make informed choices about who should represent them at all levels and this why we will continue fighting misinformation, promoting free speech and intelligent citizen debates while providing open access to information. Orodata being a non profit that’s affiliated with a tech company AI Envoy presents a combined frontier to introduce and raise these issues to the people and the government at all levels.

No Comments

Post A Comment