Harnessing the Power of AIs in Election Observation
Pic Credit: mypanhandle.com
It is now common cause that election observation has undergone profound changes over the years and this transition has been marked, in the main, by a wholesale embrace of election observation by many countries going into or holding elections. At the same time, countries are embracing new methodologies and technologies for observing elections by both domestic and international observers. This evolution is informed – in part by feedback from different election observation findings and the need to make observation of elections easy and manageable. The growing interest in Artificial Intelligence (AI), though not recent, has added another layer of intrigue to the discourse around election observation in addition to the increased use of technology and scientific approaches to the process. Increasingly, questions are being asked about the role of AIs in election observation more so the nature and depth of its role. Put somewhat differently, to what extent can AIs play a role in election observation? While some contend the process of its growing use is simply election observation automation that has been witnessed over the years, others tend to throw caution to the wind urging there are too many unknowns in its use and as such its growing use in election observation must be approached with caution.
It goes without saying the growing interest in AI has triggered debate, not only in the field of election observation but across different fields with a significant number questioning the utility and role of AI in their respective fields. From a cursory view of media reports, it appears many of the anti-AI commentaries are based on a general misconception that AI “will become so intelligent that it surpasses human intelligence and becomes our master”, while some fear being made redundant by AIs. Their reasoning seems to stem solely from what they are hearing, not necessarily researched, which on face value paints a picture that appears at odds with the traditional models of election observation. This is problematic in that such contentions may have ignored the fact AIs have been with us and used for some time in election observation. Their assumption is AIs are a recent introduction to election observation. However, such assertions present half-truths and are quite contrary to the conviction that when one takes a closer look at all the facts at play AIs have been here for some time albeit in different forms, even way before the introduction of ChatGPT AI chatbot.
AIs as Election Observation Tools
Election observation, though past infancy, is still in its developmental stages despite being in existence for decades. The concept’s methodologies and processes continue to evolve informed by different election observation findings and new observation methodologies and techniques. How do we position AIs in this evolution and how can we harness AIs to improve the quality of election observation? AI has shown its potential as a great tool for data analysis and number crunching. It has also shown and will continue to show its potential in the observation of the different phases of the electoral process such as boundary delimitation, especially when assessing data used to map boundaries, geo-positioning of boundaries and mapping of polling stations. Its potential has also been showcased when observing voter registration, more so during voter registration audit processes such as computer audits and field audits. In addition, on election day, AI tools have been used to track and analyze results both from the field and the tally centres. Similarly, they have been used as tools to audit results post-election. Election violence monitoring uses AI tools such as Ushahidi to track and report on election-related violence, while tools such as CrowdTangle have been used to monitor social media. Data analytics tools such as Tableau and Microsoft Power BI that support data visualization and analytics to create reports have proved their usefulness in election observation. Similarly, AI tools such as Time Hero have shown their potential in team management and can be a great tool for remote observer management whilst Tome has proved its usefulness as a presentation creation tool.
Are Election Observers and Experts Going to be Obsolete?
Some experts within the election observation field are worried the increased use of AIs will result in election observers and experts losing their jobs. Such fears are unfounded, based on conjecture and lack of knowledge on the use and functionality of AIs in workspaces. The reality is that human beings cannot be replaced in election observation. AI can be used for idea or content generation, but the human factor is still required. Take for example data mining, it’s given that reading and making sense from data will still require the human mind to draw inferences from the data and to contextualize it within the past and prevailing political environment in a country. The human mind will be required to read into numbers. Even before getting to the analysis, someone, a human being, must prompt the AI to provide the information required. AI will do some of the work but the human mind will get you the reading and understanding into the data and information mined by AI. Election Observer groups rely on physical field-verifiable information and data, and this can and still comes from the physically deployed observer. Jobs and workload may change but not to the magnitude of making the election experts and observers obsolete.
Cautious Experimentation and Training are Required.
The use of AI in election observation is here to stay and will change our work for the better. AIs are going to be another tool just like Excel, for example, in data analysis and other processes. The question to be asked is, how do we prepare ourselves for the inevitable increased use of AIs in election observation? The pushback for its increased use in the field is going to be there, but the transition is inevitable. Some in the field of election observation will remain skeptical about its growing utility and they are not going to be moved from their cheese (borrowed from the book, Who Moved My Cheese by Spencer Johnson). It’s given that most of the AI tools used in election observation are still in their infancy and their full potential is still to be realized. Furthermore, AI algorithms in election observation are still in their developmental stage and can be biased leading to inaccurate outcomes from their use. This can be risky in a politically sensitive field such as election observation where accurate information and data are a prerequisite. In addition, privacy issues must be considered when applying the use of AI tools in election observation, especially in fragile and closed democracies. A lot of personal data and sensitive information is gathered and stored, and its security must be ensured in case it gets into the wrong hands and may be misused for personal gain or the targeting of observers notably in fragile states. The change and transition to full-blown use of AIs in the field of election observation must be managed by gradually integrating AI into work processes. Cautious experimentation is required accompanied by training not only in its use and application but also in data literacy.
Harnessing the Power of AIs in Election Observation
Although AI has been around for some time, its use in election observation is still evolving. It is fast becoming a valuable tool for improving the quality of election observation. When observers issue positive reports, informed partly using AI tools, it helps build trust in the democratic process and enhances the legitimacy of the governments that emerge from elections. It’s, however, important to caution against its over-reliance at the expense of the human factor. AIs can facilitate faster work processes but cannot replace human problem-solving skills. They will make work in election observation easier, faster, and more efficient but cannot replace the human input into work processes.
Taona Mwanyisa, the Managing Partner of @TDGHRconsult, is a specialist in election-related support with a diverse clientele including electoral management bodies, political parties, civil society organizations, civil registries, and law enforcement agencies.
コメント