New Frontiers in Conflict Prevention: Integrating Artificial Intelligence in Early Warning and Response Systems in Kenya

New Frontiers in Conflict Prevention: Integrating Artificial Intelligence in Early Warning and Response Systems in Kenya

Solomon Kimaita, Eric J. Irungu

International Relations and Diplomacy Zetech University, Kenya

DOI : https://dx.doi.org/10.47772/IJRISS.2024.8110185

Received: 31 October 2024; Accepted: 09 November 2024; Published: 16 December 2024

ABSTRACT

With the proliferation of intra-state conflicts in Kenya, traditional methods of detecting potential conflicts and recommending timely intervention measures have proved to be largely ineffective. This paper explores how integrating Artificial Intelligence (AI) in conflict early warning systems can enhance peace and security in Kenya. It focuses on how AI can augment human capabilities in facilitating effective proactive measures in de-escalating conflicts. The paper is guided by three research objectives: first, assessing the efficacy of AI in systematically identifying indicators and patterns of conflict-triggers in Kenya; second, examining how AI can enhance real-time sharing of information to the pertinent actors for timely interventions; and third, evaluating the impact of integrating AI in Kenya’s conflict prevention initiatives. The paper is guided by the securitization theory as propounded by Ole Wæver of the Copenhagen School. The theory dispels the notion that national security issues arise naturally. Instead, the securitizing agents within the state designate select issues as threatening, consequently moving them from low to high priority security concerns, thereby necessitating concerted action. The paper adopts an explorative research design and secondary data on conflict early warning and artificial intelligence from published and unpublished sources. It finds that AI can effectively analyse vast amounts of historical, current and emerging data in order to systematically identify conflict triggers and patterns for effective conflict prevention. Moreover, it submits that AI has unmatched capacity to sustain real-time sharing of information among the relevant actors in conflict prevention infrastructure thereby guaranteeing timely responses to check the escalation of conflicts. The paper also observes that AI has the potential of enhancing Kenya’s preparedness and response capabilities thereby minimising economic, political and human losses occasioned by violent conflicts. It recommends the urgent need to revamp the existing EarlyWarning Response Systems (EWRS) by integrating AI capabilities to amplify human capabilities. In addition, the paper recommends fast-tracking of the roll-out of Information Communication Technology (ICT) infrastructure to enhance real-time sharing of information and creating a lean Infrastructure for Peace (I4P) for effective delivery. Lastly, the paper underscores the inevitability of visionary leadership to provide policy guidelines on integrating AI in conflict prevention initiatives and the general operationalization of Early Warning and Response Systems (EWRS).

Keywords—Artificial intelligence, Conflict, Conflict prevention, Early warning and response system

INTRODUCTION

Over the years, Kenya has experienced various types and levels of intra-state conflicts ranging from resource-based and environmental conflicts, electoral-based conflicts, livestock rustling, ethnic and religious clashes, organised gangs, economic and gender disparities to institutional conflicts, among others (NCIC, 2022; GoK, 2014). These mainly stem from the multi-ethnic and multi-religious composition of the society characterised by weak national identity, glaring socio-economic disparities, strong ethno-political affiliations, inequitable distribution and allocation of national resources as well as a host of environmental-related factors. The consequences of these conflicts are far reaching and leave a lasting legacy that is hard to erase. From colossal economic losses, massive loss of lives and displacements, extensive infrastructural damages, to a severely polarised society, Kenya’s economic, social and political stability remains precarious (Chunia & Ojielo, 2012).

REVIEW OF RELATED LITERATURE

Although conflicts are inevitable, violent conflicts can be avoided. Ortiz et al (2015) submit that not all conflicts are unpredictable. It is therefore possible to identify the causes and establish conflict patterns well in advance and take pre-emptive measures to contain them in good time before they escalate. Despite most of these conflicts being predictable and recurrent, and the fact that Kenya has a peace infrastructure in place – the National Steering Committee on Peacebuilding and Conflict Management (NSC) – there has been no timely, pre-emptive and structured interventions to forestall conflicts across the country. NSC has been largely incoherent, bedevilled with disjointed implementation of peace programs that are mostly ad hoc and reactionary in nature (Wambua, 2022). There is, therefore, a strong indication of the need to bolster and deploy an effective and responsive conflict prevention tool – an Early Warning and Response System (EWRS) – that can anticipate any likely conflicts. Such a tool should have a greater surveillance capacity in order to gather and analyse relevant information on possible risks and vulnerabilities, generate timely and accurate warnings that are disseminated to the pertinent actors and, lastly, the ability to generate pro-active response measures.

According to Chaves and De Cola (2017), EWRS is a set of capacities that generate and transmit timely and meaningful information that enables would-be victims – whether communities or organisations – to anticipate and take appropriate proactive and timely interventions to mitigate harm. The system should be structured to identify and analyse conflict patterns, generate alerts of conflict risks, inform decision-making and initiate timely responses to contain violent conflict. Kenya stands to benefit a lot by harnessing the advantages of technological advancements, in particular, Artificial Intelligence (AI), in promoting conflict prevention initiatives. With superior capacity for speech recognition, data gathering and summation, recognizing trends and making informed forecasts based on historical, current and emerging data, AI can augment human capacities in conflict prevention (Zuroski et al, 2023). In addition, with the ability to facilitate real-time interactions between the stakeholders, AI can enhance the capacity of EWRS by guaranteeing that relevant and updated data is readily available to support processes of monitoring, analysis, and designing of preventive measures and responses. Apparently, the vast information available on digital technologies can play a role in promoting peace.

The causes and conflict aggravating factors are intensifying in the contemporary world and the experience in Kenya is not an exception. The country faces glaring challenges in the quality of governance, biting effects of climate change, an exponential growth in the population as well as a host of other socio-economic challenges. This promises that conflicts are bound to increase and persist in the days ahead. It is on this basis that this paper seeks to assess the inevitability of integrating AI in Kenya’s peace infrastructure in order to leverage her conflict prevention initiatives.

OBJECTIVES OF THE STUDY

This study was guided by the following research objectives:

1. To assess the efficacy of AI in systematically identifying triggers and patterns of conflicts in Kenya.
2. To examine how AI can enhance real-time sharing of information to the pertinent actors for timely interventions.
3. To evaluate the impact of integrating AI in Kenya’s conflict prevention initiatives.

METHODOLOGY

This study was based on an exploratory research design, which underscores the discovery of insights and ideas. It is, therefore, best suited to study problems that are poorly investigated which in turn aids in deriving crucial themes on the subject matter. Secondly, since the design is flexible enough, it allows the researcher to adjust the direction of the study accordingly, hence making it possible to discover other interesting and pertinent issues on the problem under investigation. This lays the foundation for future research (Satpathy, B. et al 2023).

The target population for the study is 54. This is made up of the National Steering Committee on Peace Building and Conflict Management (NSC), which serves as Kenya’s conflict early warning and response unit; Uwiano Platform for Peace, a public platform that facilitates collaboration state and non-state agencies in discourses on conflict prevention; and the 47 County Peace Committees (CPCs), which help to strengthen the national peace infrastructure at the county level. The study relied on secondary data, mainly collected from relevant published and unpublished materials from key state and non-state agencies.

The study employed a purposive sampling technique. According to Nyimbili and Nyimbili (2024), this is a non-probability sampling technique that is most applicable where the researcher has a clear idea of the traits, they are keen on investigating. The units from the population are therefore selected on the ground that they bear such traits. To guarantee an adequate sample in order to draw generalised and valid inferences, the study relied on Yamane’s Formula: n = N/ (1+ N(e)2) to arrive at a sample of 17. Where n= sample size; N= population size and e= margin of error (calculated at 20%).

Table 1: Population and Sample Size

Source: Authors (2024)

THEORETICAL FRAMEWORK

This study is based on the securitization theory as propounded by Ole Wæver of the Copenhagen School. It is a departure from the traditional notion that national security issues arise naturally. Instead, the theory endeavours to explain ‘why’ and ‘how’ issues are factored into the state’s security agenda. This means that an issue becomes a security problem once the securitizing agent/actor securitizes it, that is, labels it as such. Consequently, the issue shifts from the realm of ‘normal politics’ (low security priority) to ‘emergency politics’ (high security priority), giving the state a leeway to control or combat it swiftly and, sometimes, without following the normal procedures and regulations of policy-making (Taureck, 2006).

This also provides a legitimate and justifiable ground for the state officials to take extraordinary measures, for instance, allocation of state resources, to forestall the identified issue or threat. A successful securitization process is thus made up of three steps: First, securitization move, which entails the identification of the threat. Secondly, recognizing that the elimination/containing the threat as an emergency and lastly, developing extraordinary measures to eradicate the threat (Durak, 2024).

Securitization theory finds its relevance in this study by providing the grounds for state agents to recognize and treat the causes and outcomes of armed conflicts in Kenya as existential threats. In addition, integrating AI in EWRS is a high priority security requirement in Kenya’s infrastructure for peace. Hence the theory provides a basis for prompting the state’s urgent and extra-ordinary consideration in instituting it.

RESULTS AND DISCUSSION

The study established that since independence in 1963, Kenya has experienced a number of conflicts. These range from resource-based and environmental conflicts, electoral-based conflicts, livestock rustling, ethnic and religious clashes, organised gangs, economic and gender disparities to institutional conflicts, among others (NCIC, 2022). Electoral-based conflicts are the most common and outstanding and are triggered by a set of multi-dimensional factors, among them, ethnic identity, institutional, structural and historical. Further, the conflicts are cyclic in nature, particularly since the advent of multi-party politics in 1991. The country has since witnessed violent conflicts during the general elections in 1992, 1997, 2007 and 2017 which makes it possible to draw a pattern as demonstrated in Table 2 below:

Table 2: Triggers and Patterns of Electoral-based Conflicts in Kenya – 1992 to 2017

Source: Ochieng’, et al (2023); NCIC, (2022); Wambua, (2022)

From the tabulation above, historical injustices around the distribution of land cut across as a common trigger of electoral-based violence since 1992 to 2017. Wambua (Ibid) points out that the injustices date from the colonial era and have been perpetuated by the native leaders who assumed the reins of power since independence. The suppression of grievances from the victims of land injustices by the subsequent regimes as well as the constant reference to the injustices by the political elite to advance their agenda during electioneering periods only serve to deepen the resentments. Quite often, the cumulative resentments are manifested in violent behaviour.

Institutional incapacities have always played a critical role in electoral-based conflicts. In 1992, 1997 and 2007, the Electoral Commission of Kenya (ECK) was ill-prepared to handle the general elections. The swearing-in of Mwai Kibaki as president at dusk by the Chief Justice in 2007, blatantly against the law, painted the judiciary in bad light. Even with the institutional reforms occasioned by the 2010 Constitution, there has been a general lack of public confidence in the Supreme Court, the Independent Electoral and Boundaries Commission (IEBC), the National Police Service (NPS) as well as NCIC in the execution of their mandates transparently and independently. On the same note, the media has contributed to fuelling electoral-based conflicts by spreading hate through vernacular stations. In 2007 Kass FM, a Kalenjin-based radio station, allegedly broadcast materials against the Kikuyu community while Inooro, Coro and Kameme FM, Kikuyu-based radio stations aired materials against the Luo community while painting the opposition leader, Raila Odinga, a Luo as a hooligan. In 2017, ethno-politics took a different turn, with hate spread through social media platforms.

Structural factors hinge on the inequalities occasioned by the distribution and allocation of political and economic goods. The skewed distribution of these resources in seemingly opposition strongholds heightens economic marginalisation of certain communities. Imperial presidency that quite often has annexed the legislature and the judiciary make it difficult to achieve equitability in the allocation of resources. The situation is made worse by Devolution after the promulgation of the new constitution in 2010. 2013 and 2017 were characterised by bitter contestations on revenue allocation while aggravating ethnic balkanization of the country along ethnic lines was apparent in 2017.

Ethnic undertones have as well triggered electoral conflicts. Kenya is a multi-ethnic state and, unfortunately, politics exhibits strong ethnic affiliations. Although Kenyans have coexisted peacefully, the political elite have always exploited this primordial trait to set one community against another. Ethnic discourses that characterise non-native residents of some communities in the Rift Valley, Western and Nyanza regions – ‘madoadoa’- as well ‘Wabara’ down at the Coastal regional have played a significant role in igniting violent conflicts.

Since the conflicts are predictable, cyclic and recurrent, an effective EWRS would be an opportune conflict prevention tool that can inform preparation and response actions to forestall eruption, escalation, prolongation and relapse of violent conflicts. Palau and Zambrano (2017) identify some integral components of an effective EWRS. First, risk knowledge, which entails a systematic collection and analysis of data stemming from such triggers in order to guide preparations for appropriate decision-making and response initiatives. Secondly, monitoring is at the core of an EWRS. It is the continuous surveillance of the triggers through observation of conflict parties’ behaviour, expression of the observed behaviour in numerical form (for instance, how many lives have been lost) and what is expected in the future based on this measurement (forecasting). It must, therefore, have a scientific basis in order to guarantee reliability of the forecasts and predictions.

Based on this, integrating Artificial Intelligence (AI) in EWRS would be a shot in the arm in Kenya’s conflict prevention initiatives. AI creates important innovative opportunities to increase the capacity of EWRS to collect and analyse enormous amounts of data. AI can also systematise and thus increase the accurateness and scale of EWRS as they monitor conflict dynamics and inform proactive interventions. These reinforced analyses and systems can then better promote conflict prevention by managing large-scale data and conflict monitoring to enhance timely dispatch of warnings to the relevant stakeholders. In addition, with some triggers evident in the digital platforms, AI can be useful in collecting voice, video, and data from commercial cables as well as generate predictive analytics based on social media and other types of digital information derived from repetitive human behaviour. This would be useful in understanding recurrent conflict patterns and forecast potential conflicts (Albrecht, 2023).

From the second objective, the study established that Kenya’s EWRS is housed under the Infrastructure for Peace (I4P). The I4P brings on board multiple actors who play different but interconnected roles in the conflict prevention process. However, bulk of the communication between the parties is through short message services (SMS) and email services. Thus, maintaining real-time communication remains a challenge. The problem is further compounded by poor telecommunication and internet network coverage as well as disparities in accessing appropriate communication gadgets, in most cases leading to the loss of geographical precision in data collection (Chuma & Ojielo, 2012; Wambua, 2022). The figure below shows the actors and their communication patterns.

Figure 1: Communication Patterns between Actors in Kenya’s I4P

Source: Authors, (2024)

Key:

The figure highlights the two-way communication patterns that exist between the I4P actors. At the base of the infrastructure there are the LPCs, Uwiano Platform, FBOs, WBOs and the local media outlets which play the roles of data collection as well as dissemination of warnings. Analysis of data, formulation of warnings and implementation of response actions happens at the CPCs, IOs, KMPN and ROs levels. At the top is the NSC which is the focal point of the infrastructure and is housed within the Ministry of Interior and Coordination of National Government.

The usefulness of an EWRS lies in the seamless coordination between the actors. Dissemination and communication of the warnings is therefore another integral component of an effective EARS. This entails passing clear messages containing simple information necessary to enable formulation of appropriate responses to prevent an escalation (Ibid, 2017). Therefore, there is a dire need to invest in a means that has the capacity to deliver communication to these actors on a timely basis to enable them to execute their roles in the process. AI does not only have the capacity to establish long-distance communication but also to facilitate real-time interactions between the actors. Further, to avoid burdening the actors with bulk information that may not necessarily be useful to their roles, AI can be automated to filter and customise the output in order to suit the needs of the different actors. Through automation, AI can improve real-time accuracy of data analysis at the national and local levels. With accurate analysis at their disposal, actors can easily establish conflict trends and dynamics, allowing them to formulate relevant preventive measures which in turn inform sound response capability.

In regards to the third objective, the study confirmed that most violent conflicts in Kenya are predictable and recurrent. Therefore, it is not only possible to establish the conflict trends but also to forestall their escalation or, at the very least, mitigate their negative impacts once they escalate. Unfortunately, most of the interventions are impromptu and fail the test of promptness. In addition, the study found out that there is a bulk of data to mine due to the historical nature of conflicts and the multiple institutions involved. Further, mobilisation to participate in violent conflicts in Kenya is gradually shifting from physical to digital platforms. Regrettably, Kenya’s I4P relies heavily on human capacity to collect relevant data, analyse and relay it to the pertinent actors. Overall, these factors limit the timely dissemination of information to the relevant actors and, consequently, hampers response mechanisms. As a result of these inadequacies, Kenya suffers losses that could have otherwise been avoided. The table below summarises the impacts in selected sectors.

Table 3: Impacts of Violent Conflicts in Kenya

Source: Authors, 2024.

The impacts of violent conflicts fall heavily on the different sectors. With a poor economy, Kenya is not able to sustain post-conflict reconstruction on the damaged infrastructure. This creates a conducive environment for conflict relapses. Constant foreign direct investment flights, decreased workforce due to deaths and replacements as well as increased security spending contribute to underdevelopment. Politically, democratic principles are greatly undermined which makes the country susceptible to instability. From a social perspective, co-existence among communities is characterised with suspicion while involuntary displacements give rise to multiple humanitarian crises. The agricultural sector is a major employer in Kenya as well as a source of food. With constant conflicts, the country risks food insecurity and massive loss of livelihoods. The most notable impacts on the health and education sectors are increased medical costs and high school drop-outs, respectively, and massive exodus of the professionals. Lastly, violent conflicts make it difficult to uphold the rule of law and a soaring record of human rights violations from different quarters.

Although integrating AI in Kenya’s EWRS is not a panacea to conflict prevention, it can greatly augment human capacity in mitigating violent conflicts and, consequently, their far-reaching impacts. AI has the capacity to automate data extraction and analysis – including data available in digital platforms – generate accurate predictions and aid actors in decision-making (Albrecht, 2023). Palau & Zambrano (2017) hold that the anticipation of an occurrence permits well-reasoned and timely decision-making, allowing ample time to formulate, communicate and implement responses hence reducing the extent of destruction and/or harm. Since Kenya has a database on the conflict triggers and some established conflict patterns (although disjointed), Zuroski, et al submit that it is possible for AI to categorise such data and make informed predictions based on new and emerging data. This would be a big reap for Kenya’s conflict prevention initiatives culminating in the reduction of negative impacts in the various sectors outlined in Table 3 above.

CONCLUSION

It is not contested that Kenya has an ambitious I4P in place. However, the infrastructure has not lived up to its potential, at least from the number of recurrent conflicts witnessed in the country, even after its inception in 2001. The study observed that the operations of the I4P are, to a large extent, manual, disjointed and ad hoc. As a result, the country experiences fragile peace and is vulnerable to conflict escalations which could have otherwise been contained in good time. Integration of AI in the I4P is inevitable in order to seal these loopholes. There is therefore an urgent need to revamp the EWRS in Kenya by investing in the capabilities of AI as an amplifier of conflict prevention initiatives. This will guarantee systematic identification of triggers and patterns of conflicts, timely development and implementation of response mechanisms and eventually avert violent conflicts.

Secondly, the study observed that there are multiple actors in Kenya’s I4P and communication between them is largely asynchronous. This is partly contributed by poor information communication technology network and disparities in access to requisite gadgets by lower-level actors. It is also observed that the I4P is bloated. Equipping the infrastructure with cutting edge technology would, most probably, strain the government economically. Investing in AI powered EWRS under such circumstances would be underwhelming. Thus, it is incumbent upon the Kenyan government to fast track the roll-out of the ICT infrastructure across the country. This is necessary to ensure real-time exchange of information among the relevant actors for timely action and strengthening of the teamwork thereof. On the same note, maintaining a lean but well-structured I4P would not only make it possible for the government to fund and equip it but also reduce the risk of duplication of mandates.

Lastly, conflicts have a lasting legacy. It is unfortunate that Kenya has to contend with the consequences of violent conflicts that could have been easily prevented. This rests squarely with the quality of leadership and governance in the country. Revamping the EWRS by integrating AI, among other policy guidelines, does not guarantee the successful conflict prevention. The success of EWRS is largely dependent on visionary leadership that recognizes the benefits of investing in a responsive EWRS. This would instil public confidence and legitimacy in the system hence improving its efficacy.

RECOMMENDATIONS

The paper, therefore, recommends development of clear policy guidelines on how to integrate AI in the system as well as guaranteeing the independence and funding of the system. This will contribute in realising the benefits of integrating AI in conflict prevention initiatives in Kenya.

REFERENCES

1. Albrecht, E. (2023) Predictive Technologies in Conflict Prevention: Practical and Policy Considerations for the Multilateral System Discussion Paper. United Nations University Centre for Policy Research.
2. Chaves, J. M. & De Cola, T. (2017) Public Warning Applications: Requirements and Examples in Wireless Public Safety Networks.
3. Durak, G. (2024). Contribution of Copenhagen School to the Security Studies. Urdu Universities Sosyal Bilimler Enstitüsü Sosyal Bilimler Araştırmaları Dergisi, 14(1), 336-350.
4. Espen, G. R. et al (2024) A Review and Comparison of Conflict Early Warning Systems in International Journal of Forecasting No. 40, pp. 96–112
5. Government of Kenya (GoK) (2014), Sessional Paper No, 5 of 2014 on National Policy for Peacebuilding and Conflict Management.
6. International Strategy for Disaster Reduction (ISDR) (2006) Developing Early Warning Systems: A Checklist.
7. National Cohesion and Integration Commission (NCIC), (2022) Towards a Violence Free 2022 Elections: Conflict Hotspot Mapping for Kenya.
8. Nyimbili F. and Nyimbili L. (2024) Types of Purposive Sampling Techniques with Their Examples and Application in Qualitative Research Studies in British Journal of Multidisciplinary and Advanced Studies: Vol.
9. Satpathy, B. et al (2023) An Introduction to Research Methods and Designs in Social Sciences in Advances in Agricultural Research Methodology. ResearchGate.
10. Taureck, R. (2006). Securitization Theory and Securitization Studies in Journal of International Relations and Development, (9) pp
11. Ochieng’, C. O. P. et al (2023) Causes and Consequences of Post-Election Violence in Kenya in African Journal of Empirical Research Vol. 4 No. 2, pp. 480-492
12. Ojielo, O. and Chuma, A. (2012) Building a Standing National Capacity for Conflict Prevention and Resolution in Kenya in Journal of Peacebuilding & Development Vol. 7, No. 3, pp. 25-39 Sage Publications, Inc.
13. Prieto-Remón, T. C., Cobo-Benita, J. R., Ortiz-Marcos, I., & Uruburu, A. (2015). Conflict resolution to project performance. Procedia-Social and Behavioural Sciences, 194, 155-164.
14. Wambua, M. (2022) Strengthening Kenya’s Infrastructure for Peace towards 2022 Elections (2017) The Ethnicization of Electoral Conflicts in Kenya: Options for Positive Peace. ACCORD.
15. Zambrano, A. M. and Palau, C. (2017) Community Early Warning Systems in Wireless Public Safety Networks.
16. Zuroski, N. et al (2023) Designing AI for Conflict Prevention & Peacebuilding. UNSC, New York.