Building Bridges with AI

How the UN is using artificial intelligence and algorithms to facilitate more inclusive peace negotiations.

When you think of customer research, you don’t usually think of peacekeeping. Just as algorithms are often viewed as proprietary “black boxes”, conflict negotiations thrive in the furtive darkness of back channels and secure situation rooms. Rarely are citizens encouraged to voice their opinions on deals or share insights from the frontlines. The UN is looking to change this, with help from an innovative AI company called Remesh.

Remesh is an interactive platform that uses algorithms to help solicit and analyze opinions at scale. Fueled by natural language processing, Remesh both facilitates and mediates real-time conversations with large groups of people online, producing sentiment analysis on key themes and insights surfaced from the crowd. To date, the platform has typically been used by large companies or research firms, looking to better understand anything from brand awareness to product performance. Applying artificial intelligence to customer conversations is a huge step forward for many corporations. Tools such as focus groups and surveys have seen little innovation and can be ineffective ways to gauge customer needs and perceptions.

As it turns out, innovations in conversation data and analytics can also be a game changer when it comes to crafting effective peace treaties. In an effort to better understand the needs of citizens living in conflict zones, the UN is working with Remesh to run smartphone polls to collect opinions. The insights generated from these activities would be relayed back to negotiators and policy analysts, who would then be able to amend deals to ensure that everyone has a voice in the mediation. Not only is this a more inclusive way to approach peacekeeping, but it also might be more effective; studies show that wider consultation is the key to creating lasting peace and avoiding conflict resurgence.

Of course, there are many disadvantages to introducing artificial intelligence and algorithms into conflict negotiations. While this is undoubtedly a step forward in terms of transparency in the already opaque peacekeeping process, it does introduce the possibility of bias in the negotiation. Since the conversations are only conducted over a smartphone, access is a concern. The UN says it hopes to combat this issue by reaching out to potential users with both online and physical invitations to participate. The potential for bias still exists however, as the platform still favors those with both the desire and capacity to engage.

More labor-intensive methods such as field research and in-person surveys would ensure a uniform collection of insights, as people without the time or resources to access mobile polls would still be included in the discussion. However, the difficulty in engaging with citizens, particularly in conflict zones, cannot be underestimated. In addition to the time and workforce that it would take to gather enough opinions from the population to surface common sentiments, manual data collection would also slow the peace process overall and extend out the time to resolution. Online collection powered by algorithms allows the UN to collect and analyze opinions quickly and adjust to any changes in conditions. Indeed, any new proposals or amendments can easily be sent out for assessment and immediate feedback. The flexibility of the Remesh platform matches the fluidity of an evolving conflict situation and ensures that any surfaced sentiments remain reflective of current conditions.

While it remains to be seen how useful this tool is in resolving conflicts, it is exciting to see the UN embrace AI and sentiment analysis in an effort to include more voices in peacekeeping negotiations. The partnership also shows us that innovations in harvesting and analyzing opinions can be broadly applied, transforming more than just the domain of corporate marketers and pollsters. Conversation analytics and algorithms have the capacity to change how we communicate and will continue to serve as powerful tools that unlock how we develop and share beliefs.


HireVue: A face-scanning algorithm decides whether you deserve the job


Organizational network analysis: overcoming common pitfalls

Student comments on Building Bridges with AI

  1. Thanks for sharing! I believe that in general, the more data and opinions one can collect prior to developing a strategy, the better. However, as you pointed out, there are several issues that one must consider before drawing insights from a dataset. In this particular situation, I agree with you in that my main two concerns are: access to the opportunity to participate in the survey and participation bias. Given the potential impact of the conclusions drawn from this data, I would also like to see demographic data tied to opinions. I feel like doing this can help mitigate against the risk of a political group cherry-picking a sample set and projecting the results onto a population.

Leave a comment