AI can get more young South Africans to turn out for the elections

Education, information, reminders and fact checking all available at a fraction of the cost of traditional methods

With voter registration officially closed following the announcement of the election date, the Electoral Commission of SA (IEC) reports that the voters roll stands at 27.8-million. Young adults (18-29) make up a paltry 18% of that figure, reflecting a general lack of interest in electoral participation among the youth compared with the situation as recently as the 2019 election. Solutions are required on multiple fronts to arrest this decline, and it is incumbent on the government, civil society and the private sector to tap into the rapid advances made in artificial intelligence (AI), which offer immense opportunities if applied appropriately.

While SA’s electoral machinery has various constituents, the overall objective of each participant is a free and fair election. Citizens want to feel that they can trust the electoral process and infrastructure; political representatives seek an efficient electoral commission that ensures credible and trustworthy outcomes; and the media’s role is to report in an ethical manner that empowers citizens to make informed decisions.

The advent of generative AI, specifically chat generative pre-trained transformer (ChatGPT) —

AI that acts in a conversational way — could be applied at various phases of the build-up to May 29, from the recently concluded preregistration phase to the pre-voting phase leading up to the actual day of voting and beyond.

The promise of AI is that it can generate content in response to prompts using natural language in various formats, such as voice and text — in this case election-related education and information, quickly and at a fraction of the cost of traditional methods. Since those providing voter education — from the IEC to NGOs — have limited budgets, this means AI can solve some of these constraints.

At preregistration, the deployment of an app with an AI voting assistant to produce more compelling content might have yielded a higher voter registration, specifically among the youth. This assertion holds weight when one considers that a significant percentage of 18- to 26-year-olds check their cellphones at least 30 times an hour. This equates to spending at least a quarter of every day engaged with their cellphones.

The use case of AI here is apparent — legitimate stakeholders such as government, political parties, businesses and academic institutions can leverage off natural language-based communications and personalisation at scale, to create content for their youth constituents to understand their civic duty about the voting environment.

Awareness campaigns can also be uniquely tailored to their segment, language and demographic. These AI marketing techniques would yield measurable outcomes that could reduce marketing costs and, more importantly, result in higher voter turnout among the youth.

Such a voting assistant can also provide comprehensive civic education around voting policies and manifestos and offer simulations of what the future could look like based on scenarios during the pre-voting phase.

Existing applications such as Vote Compass, Election Buddy and Mpesa offer useful case studies in this regard. In addition to robustly marketing its voter application among the youth, the IEC could use the app to take the conversations to where the youth are. It would enable interactive dialogue to promote inclusion and get them actively involved, in addition to sending through registration reminders.

Though shifting voter registration among the youth with the help of AI would be a significant step, more work would still be required in the lead-up to May 29. In the 2019 general election, of the 19% of eligible 18- to 19-year-olds who registered only 15% turned up at the polls. In the 20- to 29-year-old age group 30% of those who registered voted. This points to an urgent need to ramp up efforts to encourage voter participation among the youth, or a whole generation may not get to participate in achieving democratic outcomes.

In addition to voter participation it is important to explore the use of AI to aggressively counter misinformation and disinformation — fake news. Consider the Oxford Institute’s Global Inventory of Organised Social Media Manipulation study, which identified 77 countries where government or political party actors used disinformation on social media to manipulate public opinion.

The report found that in SA government agencies, politicians, political parties, private contractors, citizens and influencers are involved in social media manipulation. Two years later, Media Monitoring Africa director William Bird told the SA Human Rights Commission that the disinformation posts shared during the July unrest in 2021 sowed anger and contributed to the widespread violence and looting.

Nigeria offers a useful case study of how AI can be used to counter this threat. In the lead-up to the February 25 election last year Nigerian fact checkers used new AI tools, developed by Full Fact with support from Google, to fight bad information. As part of the Nigerian Fact-Checkers Coalition the fact checkers were able to use these tools to debunk common election disinformation trends, and fact checked false claims attributed to politicians and past presidents.

The extent of AI’s effect on elections and voter participation was most recently felt in Indonesia, a nation where almost half of the voters are under the age of 40. In that country generative AI was used to rebrand one of the presidential candidates by using a “cuddly grandpa avatar” in the lead-up to the February 14 elections. Political consultants in the country are already using apps such as Pemilu.AI, which uses OpenAI’s GPT-4 and 3.5 software to craft hyperlocal campaign strategies and speeches.

To encourage co-ordinated stakeholder participation in using AI to scale voter participation and counter misinformation the SA government must take deliberate action to strengthen its AI policy environment and build in sufficient safeguards through multilateral efforts such as the landmark Bletchley Declaration, which is aimed at boosting global efforts to cooperate on AI safety.