Utilizing AI to Improve Efficiency of the Environment and Land Court in the Kenyan Judiciary Leveraging AI Capabilities in Land Dispute Cases in the Kenyan Environmental and Land Court System Florence Ogonjo Joseph Theuri Gitonga Dr. Angeline Wairegi† CIPIT CIPIT CIPIT Strathmore University Strathmore University Strathmore University Nairobi, Kenya Nairobi, Kenya Nairobi, Kenya fogonjo@strathmore.edu jgitonga@strathmore.edu awairegi@strathmore.edu Dr. Isaac Rutenberg CIPIT Strathmore University Nairobi, Kenya irutenberg@strathmore.edu ABSTRACT and poverty that continues to plague the nation [2, 3]. The transition from pre-colonial communal land ownership to private land The number of land disputes in Kenya continues to increase with ownership, which started in the colonial period and has continued population and economic growth. In 2013, the judiciary established in the post-colonial era, produced a number of contradictions in the Environment and Land Court (ELC) to hear disputes relating to administering and managing land that are present to this day [4]. A environment and land. Unfortunately, the ELC is plagued with the large number of land policies enacted in the colonial era persisted same problems affecting Kenya’s other courts; chief amongst these even after the country attained its independence in 1963. As a is an extensive backlog of cases. Past attempts by the judiciary to result, despite enacting individual tenure of indigenous land and eliminate this backlog have met with varying degrees of success. In redistributing the fertile lands in the highlands occupied by this paper, we argue that augmenting human abilities with AI colonialists to its citizens, Kenya is plagued with land conflicts and technology is a viable means of tackling this case backlog. This its courts inundated with land dispute cases among individuals and paper outlines AI tools that may aid legal personnel in the ELC in between communities [5]. performing their duties and, ultimately, reducing the number of Changing cultural practices are increasing pressure on the pending cases. country’s land tenure system as well. For example, the Pokot tribe in the highlands of Baringo county traditionally practiced semi‐ KEYWORDS nomadic pastoralism; in recent years, however, the community has Kenyan Judiciary, Environment and Land Court, Artificial adopted a more sedentary lifestyle and taken up rain‐fed agriculture Intelligence (AI), Legal Research, Transcription, Online Dispute [6]. The transition from common property to private tenure has led Resolution to increased land disputes within members of the tribe [6]. In the past, people accessed land and asserted their land rights in 3 ways: (i) using clan-based definitions of landholding communities, (ii) 1 INTRODUCTION through family-based inheritance, and (iii) from claims to rights Kenya’s relationship with land can be viewed in pre and post – based on long-term occupancy and use [1]. Tensions between these colonial terms: a precolonial context of land abundance and relative traditional land administration methods and current land laws are labor scarcity, and a late colonial and postcolonial situation of also contributing to land disputes in the nation. The Maasai’s rising populations and growing pressure on land [1]. Colonial land customary land holding, for example, is based on long occupation, policy in Kenya resulted in inequality in land ownership and use, continuous use, traditional rights, colonial treaty, and the Group resentment by Africans, landlessness, squatting, land degradation Representatives Land Act adopted in the early years after independence [7]. In recent years, the tribe’s people are † Corresponding author. experiencing dispossession from people making claims to their land In: Proceedings of the Second International Workshop on AI and Intelligent based on the more formal land laws currently in place [7]. Assistance for Legal Professionals in the Digital Workplace (LegalAIIA 2021), held in conjunction with ICAIL 2021, June 21, 2021, Sao Paulo, Brazil Gender disparities in land ownership and access to land are Copyright © 2021 for this paper by its authors. Use permitted under Creative prevalent in Kenya. Women lag behind in securing land rights. Commons License Attribution 4.0 International (CC BY 4.0). Published at http://ceur-ws.org Women, particularly those residing in rural areas, are more likely to be systematically excluded from family and patriarchal land LegalAIIA’21, June, 2021, Sao Paolo, Brazil F. Ogonjo et al. ownership – only having access to land through a male relative – influenced by the backlog of land cases in the High Court. Case leaving them in precarious financial situations [8]. This can lead to backlog - defined as cases that have been pending for more than land disputes. A 2005 study found that widows in the country are one year - and delays in delivery of justice have been main 13% more likely to experience land conflicts when their parcels are indictments against the Kenyan judiciary. registered under the names of their deceased husbands than when Underfunding of the judiciary is one of the main reasons for the titles are registered under their names [9]. increasing case backlog in courts across the country. The entire Generally, the source of land conflicts can be grouped into 3 judiciary system is regularly subjected to abrupt budget cuts from categories: (i) competing land claims from agricultural and the government; in fact, the Judiciary’s funding has been on a urbanization demands as a result of rural – urban migratory patterns downward trend since the 2012 - 2013 fiscal year [20, 21]. [10], (ii) conflict between the elite and ordinary citizens in terms of Budgetary constraints mean that the judiciary is consistently land distribution, natural resource extraction and unbalanced rent operating with less than the required workforce (55% of the sharing [11, 12, 13, 14, 15], and (iii) land grabbing using political required workforce in the 2019 – 2020 fiscal year [21]) which or civil machinations [14, 16, 17, 18]. The number of disputes over inevitably results in an increase in the number of pending cases. For land continue to increase with population and economic growth. As example, in the 2016 – 2017 fiscal year there were 499,341 pending family patriarchs that acquired land in the early Independence era cases, and this number increased to 617,582 by 2019 – 2020 [20, pass there is an increase in land inheritance disputes which also 21]. Other factors that contribute to the backlog include: poor contributes to rising land conflicts. physical infrastructure – there has been minimal progress in the Prior to 2010, Kenya did not have a comprehensive system of completion of court construction across the country; lack of ICT land laws, particularly those pertaining to women’s land rights. In capacity to assist with core judicial processes; judicial 2010, it adopted a new constitution that allowed for the organizational structure; court rules and procedures; and manual implementation of the National Land Policy through institutions management of court records [20, 22]. such as the National Land Commission Act (NLC). The It is worth noting that the judiciary is actively working on formulation of a comprehensive National Land Policy commenced reducing case backlog. In the 2018 – 2019 fiscal year, under Chief in February 2004 and was completed in 2009. The National Land Justice David Kenani Maraga, the judiciary significantly reduced Policy was commissioned by the government to tackle issues of the number of cases pending in the court system - bring down cases squatting, landlessness, disinheritance of some groups and that are five years old from 110,000 to only 15,278 cases [23]. The individuals, urban squalor, under-utilization and abandonment of issue of case backlog is a perennial headache for the judiciary, agricultural land, deterioration in land quality, tenure insecurity, though, with Chief Justice Maraga noting that the number of cases and conflict [8]. The National Land Policy recommended the filed in Kenyan courts every year exceeds the number of cases the creation of mechanisms to ensure access to timely, efficient and judiciary settles by as much as 100,000, causing an ever-growing affordable dispute resolution to land conflicts. This led to the backlog [24]. establishment of the Environment and Land Court. The ELC has one of the highest number of pending cases, Following promulgation of the 2010 constitution, an 13,630 cases for the 2019 – 2020 fiscal year, across all counties Environment and Land Court Act was approved in 2011 under despite an impressive case clearance rate [21]. Of the 13,630 cases, which a new Environment and Land Court (ELC) was created as a 21%, 2920 cases, have been in the court system for greater than 5 specialized court. Currently, 29 of the country’s 47 counties have years, 34%, 4628 cases, are aged between 3 – 5 years, and 45%, an ELC. The Environment and Land Court is a Superior Court with 6082 cases, have been pending for 1 – 3 years [21]. Most of the the same status as the High Court of Kenya. It has jurisdiction to disputes in the ELC are family disagreements over land and fraud hear disputes relating to environment and land. Specifically, the related cases, 29%; succession cases account for 20%, boundary court has the power to: (i) hear disputes relating to land disputes, 15%, and double registration and double allocation administration and management, (ii) hear cases relating to public, accounted for 10% and 9% of the cases, respectively [2]. The private and community land and contracts, (iii) hear cases relating impact of these protracted legal proceedings is devastating to the to environmental planning and protection (iv) exercise appellate parties involved in the disputes. jurisdiction over the decisions of subordinate courts or local tribunals, and (v) exercise supervisory jurisdiction over the subordinate courts, local tribunals, persons or authorities [19]. The ELC began its operations in 2013. The majority of the cases handled by ELC relate to land disputes rather than environmental matters. 2 CURRENT PERFORMANCE OF THE ENVIRONMENT AND LAND COURT Unfortunately, the ELC is plagued with many of the same problems affecting Kenyan courts; chief amongst these is an extensive back Figure 1: The total number of pending cases over a year old log of cases. Ironically, establishment of the ELC was largely across the Kenyan judiciary system as reported in the State of Utilizing AI to Improve Efficiency of the Environment and Land LegalAIIA’21, June, 2021, Sao Paolo, Brazil Court in the Kenyan Judiciary the Judiciary and the Administration of Justice Annual Report 2019 – 2020 [21]. In order to accurately assess the current performance of the ELC this study evaluated the total number of pending cases and their average age in all 29 ELC county stations. It also examined the average age of cases in terms of those that were ultimately dismissed and those allowed to proceed through the court. This data was obtained by analyzing court fillings in the Environment and Land Court posted on the Kenya Law online website, the official account of the Kenyan judiciary. A computational search of the court case meta-data was performed and the relevant data compiled. Between 2002 – 2020 the number of land cases filed in Environment and Land Courts increased significantly. The distribution of the number of cases filed each year is shown in Figure. 2. The significant decrease in the number of cases in 2020 was due to a lack of availability in the court’s calendar, i.e., the case backlog prevented parties from scheduling new hearings in the ELC. Chief Justice David Maraga announced in January 23, 2020 that the Environment and Land Courts are fully booked until March 2021, and called for urgent measures to fund the Judiciary’s bid to employ more judges and magistrates to deal with the case backlog [25]. Figure 3: A mapping of the number of pending land cases filed in Environment and Land Courts in counties across Kenya. The highest caseloads are in Nairobi, Nakuru and Meru counties, with 18%, 7% and 7% of the 14,686 total pending cases, respectively. The average age of these cases is 3 years. The highest average durations from filing to judgement are in Meru county, 4.5 years, Machakos county, 4.3 years and Bungoma county, 3.8 years. Figure. 4 outlines the average duration of cases in the ELC system in counties across the country. It is clear that the current court system is ill equipped to deal with the demand of cases. In the following section, we argue that AI tools could be used to significantly decrease, or eliminate, the Figure 2: The total number of land cases filed in Environment backlog of cases. We outline the different AI tools that the Kenyan and Land Courts in 29 counties for the 2002 -2020 period. The Judiciary should introduce to the ELC to aid the legal workforce to cases between 2002 – 2012 are land cases carried over to the efficiently perform their duties and, ultimately, tackle the large ELC from the High Court. number of pending cases. There are a total of 14,686 pending land cases. The distribution 3 UTILIZING AI IN THE ENVIRONMENT AND of cases across the country is shown in Figure. 3. A total of 127 LAND COURT judges are assigned to these cases, equating to an average of 115 The use of AI in legal proceedings is not as prevalent in Africa as pending cases per judge. it is in the USA or Europe. A few African firms - Bowman, with offices in Nairobi, Kenya; Webber–Wetzel, headquartered in Johannesburg, South Africa, and KTA Advocates (formerly Karuhanga, Tabaro & Associates) in Uganda, for instance - have adopted AI to improve their legal services delivery; streamlining the mundane, time consuming tasks through the use of AI systems and freeing up their lawyers’ time to focus on high level tasks [26]. LegalAIIA’21, June, 2021, Sao Paolo, Brazil F. Ogonjo et al. and (iv) lack of funding and labor to augment legal proceedings with AI technology. This paper argues that AI holds great potential in increasing the efficiency of the Kenyan courts thereby reducing the current and future case backlog. 3.1 AI in Legal Research for Land Dispute Cases The Kenyan legal system, like many globally, is based on precedent - judges make rulings consistent with prior cases on the same subject. Judges must therefore identify and retrieve information from relevant cases to support their decision-making. The high number, and high complexity, of cases that judges must sift through makes this aspect of their job highly time consuming and contributes to the length of a trial. AI tools that aid with legal research would ease this aspect of judges’ workload. AI legal research platforms are computer software systems that not only perform pre – programmed tasks but have the ability to learn and refine their searches and outputs. Machine learning (ML) and Natural Language Processing (NLP) may offer affordable ways to obtain precise and relevant legal research results [34, 35, 36, 37]. Some of the commercially available AI legal research platforms use natural language processing to search and process data using pre- defined parameters. NLP uses prior users’ queries and results to form a predictive model, expanding or narrowing a search to ensure all relevant cases are identified. The efficacy of these tools is Figure 4: A mapping of the average age of land cases filed in documented. A 2018 study, for example, found that attorneys who ELC in counties across Kenya. The average age for pending used AI tools to conduct legal research completed projects 24.5% cases in the ELC system country-wide is 3 years. faster and the search results were 21% more relevant; the study concluded that use of AI would save attorneys 132 – 210 hours a LawPavillion, a Nigerian legal technology company, launched an year when conducting legal research [38]. AI platform in 2016, LawPavillionPrime, that gives in-depth The cost reduction potential of these AI legal research analysis of the strengths and weaknesses of legal positions and platforms should also motivate the ELC to integrate them to the authorities by generating statistical analysis, historical data, current adjudicating process. Understanding the exact fiscal impact precedential value ratings, conflicting judgments, locus classicus, of AI solutions in the ELC before implementation, however, is statutory or literary authorities, and opinions [27]. This was the first difficult. The cost of software, implementation, training and such platform launched on the continent. In 2018, it released TIMI, staffing comparative to current processing and personnel cost may Nigeria’s first artificial intelligence legal assistant, which assists not incentivize utilization of these AI systems if it is prohibitively lawyers with legal research, litigation, opinion drafting, provides higher. The previous upgrade to provide internet and Wi-Fi access notes with legal authorities, and gives a step-by-step guide on in 90% of the courts cost KSH 40 Million (£300,000) and stalled drafting and filing court processes [28]. shortly after surpassing that sum due to lack of funds [39]. If the In general, however, law offices and courts on the continent cost of deploying AI legal research platforms is similar than the have been slow to embrace technology. In 2018, for example, the same problem may arise. Fortunately, there are several AI legal Law Society of Kenya (LSK) went to court to oppose a decision by research platforms, also known as ‘Due Diligence’ platforms the Ministry of Lands and Physical Planning to digitize the land currently on the market: Kira Systems, Leverton, eBrevia, Ross transactions processes at the land registry through the National Intelligence, CaseText, WhatSun, TIMI, and many more, at varying Land Information Management System (NLIMS), arguing that the price points depending on the functionality and tools available. ministry had failed to consult the relevant stakeholders as required Subscriptions can be as low as $59 per month. [29, 30, 31]. The government argues that digitizing land registration Effective deployment of these tools will also require extensive documents will root out corruption in land transactions while the personnel and algorithm training. The cost of both of these training LSK argues that done without appropriate legislation, digitization is likely to vary. However, these initial costs are likely to be offset is likely to increase corruption in land management [32]. in the long term by a reduction in the personnel required to operate Furthermore, it is only in recent years that digital signatures and the ELC efficiently. The hiring practices in the ELC will need to be service of pleadings via email have taken root in the country [33]. drastically altered if these tools are adopted. There will be no need Currently, AI is not utilized in the Kenyan judicial system in a to recruit armies of young lawyers to perform services that are no substantive manner. There are several reasons for this: (i) poor longer needed, instead the ELC will need to hire a smaller number digital infrastructure and data capacity, (ii) under-digitization of of legal personnel adept at utilizing AI legal research platforms. records, (iii) tradition-bound court systems and legal professionals, Utilizing AI to Improve Efficiency of the Environment and Land LegalAIIA’21, June, 2021, Sao Paolo, Brazil Court in the Kenyan Judiciary These platforms are only as strong as the data they have access corporations are hesitant to invest in digital infrastructure in these to. Access to comprehensive, robust case data increases the rural communities. In order to provide digital infrastructure capable efficiency and accuracy of AI platform searches. Opportunely, in of supporting all technologies, the Kenyan government and private 2017, the Kenyan judiciary unveiled its digital strategy, enshrined industries may have to work together. The government may in the 2017-2021 Sustaining the Judiciary Transformation subsidize some of the costs of construction incentivizing private Blueprint, to re-engineer its processes through information and industries to provide better digital services in rural areas. communications technology (ICT) [40]. Part of the strategy Alternatively, both parties can opt to share the infrastructure to cut outlined is the digitization of court records and proceedings, down cost. retiring archaic filing systems and modernizing document management [40]. According to the Ministry of ICT, 60 million 3.2 AI for Speech Recognition and Transcription in ELC records were digitized under the High Court Registry pilot The official records of courtroom proceedings are vital in the digitization project [41]. There are minimal reports on the progress justice system. Legal transcription is therefore a vital component of of the digitization project in other courts. Digitization of these the adjudication process. Court transcripts influence, “…the records will make the use of AI to conduct legal research a viable conduct of the trial, whether by court alone or by court and jury; strategy. the relationships between the trial judge and participating counsel; Access to land records is necessary for effective deployment of the procedures for review of the trial by the trial judge; and these platforms. This means that complete digitization of land appellate review, including the feasibility of seeking such review records in the country is required. In April 2021, the government and the nature, scope and potential achievements thereof” [46]. launched a new National Land Information Management System Unfortunately, underfunding of the judiciary affects the number of (NLIMS), a digital land resource management platform named available court transcriptionists. This shortage of transcriptionists Ardhisasa; another step in the government’s goal to digitize land has left many courts on their own when it comes to obtaining records and transactions [42]. The phased roll out of the Ardhisasa accurate transcripts of courtroom events. Many court proceedings platform started in Nairobi (where digitalization of all services is in Kenya exist only as audio recordings. In 2019, the Judiciary and complete), with another twenty counties to be on-boarded to the the Ministry of Information vowed to digitize all audio court digital system by the end 2021. The platform is expected to be proceedings using the Ajira Digital Program, while employing available across the country by the end of 2022 – a goal that falls youth to perform the transcription [47]. This is a worthwhile but short of the 2021 completion date of digitization of land records set slow endeavor. During the COVID-19 pandemic, many Kenyan by the government in 2019. A partial digitization of these records courts were forced to adopt real – time transcript devices. Data would invariably affect the efficacy of the AI legal research tools. regarding how many, if any, Environment and Land Courts adopted Digitization of land services is likely to be more cost effective these devices could not be found. as well. In 2009, the average cost of managing the manual land AI can reduce case backlog at the ELC by filling in the gaps system was KSH 1,770.00 per file documents; it cost KSH caused by the shortage of court transcriptionists. AI coupled with 10,621.00 on average to trace a misfiled or missing documents and automatic speech recognition (ASR) allows for proceedings to be KSH 19,473.00 to reproduce a lost file or document in the land recorded, processed, and transcribed faster than using traditional registry [43]. These costs were significantly lower in developing court transcriptionists. Generally, ASR in targeted applications countries that had fully digitized their operations [43]. (e.g., legal or medical transcriptions) tends to have lower accuracy It should be noted that the country’s poor digital infrastructure, than in general purpose applications (e.g., regular speech, internet i.e., lack of internet access, poor internet connectivity, and cost search engines) [48]. However, automated speech recognition prohibitive internet, may hamper the use of these AI platforms in (ASR) technology combined with AI improves speech-to-text Environment and Land Courts located in rural counties. There is an engines increasing their ability and allowing them to transcribe ongoing push by the Ministry of ICT to improve access to high- jargon-heavy legal proceedings highly accurately [49]. In general, speed internet in rural locations [44]. However, there are number of the most effective application of these AI transcription tools obstacles hindering this goal. First, rural communities often lack augment the automated process with human oversight; the reliable electricity which makes it harder for technology companies automatically produced transcripts are reviewed and edited by to set up internet networks. The quality of available digital professional transcribers to ensure the highest level of accuracy infrastructure is also cause for concern. For example, only 57% of [49]. This would be especially critical in this context since ASR is the population receives 4G coverage in Kenya, and the majority of less accurate when dealing with accented speech [48, 50]. places not covered are rural [45]; as a result, even the simplest Corrections or enhancements are fed back to the ASR via adaptive technologies often don’t work as expected in these areas. Finally, algorithms, allowing the technology to constantly improve [49]. A the depressed income of many residents in the rural areas mean review of a transcript should take much less time than manually competing basic needs often impact the ability to access digital transcribing audio recording of court proceedings. services. Perversely, the cost of offering internet services is often Additionally, AI transcription service systems provide high higher in rural areas due to greater costs in building, servicing and searchable features, allowing for targeted data to be easily even fueling those networks [45]. As a result of lower demand, identified using relevant keywords and dispersed files to be LegalAIIA’21, June, 2021, Sao Paolo, Brazil F. Ogonjo et al. consolidated in the form of an organized digital database [51]. This dismissed were pending for a longer duration in most of the would further streamline case management flow in the ELC and counties surveyed. mitigate case backlog. Funding is both an obstacle and motivation in deploying AI Predicting judicial matters is an ongoing and longstanding transcription platforms in the ELC. In the USA, the most commonly preoccupation in legal circles that continues to be an open issue in recommended AI services cost around 25¢ per minute of audio, and both the theory practice of the law [54, 55, 56, 57, 58]. In recent services employing human transcriptionists cost up to $2 for a years, AI based approaches have been increasingly utilized for legal minute of clear audio [52]. One transcription service operating in predictive analysis. AI can be used to identify patterns in a judges’ Kenya offers rates of $1.00 per minute for legal transcription with rulings, allowing lawyers and other legal professionals to predict 5 - day delivery [53]. Therefore, in addition to increasing the speed how the court may rule. Algorithms and machine-learning can of the transcription process, use of AI transcription platforms would interpret data and predict a logical outcome for a case before filling. save the perennially underfunded ELC money. It is worth noting Environment and Land Courts publish case details including that during the 2019 fiscal year, the Directorate of the ICT judgements online making big data analysis possible. developed specifications for the procurement of a speech to text Researchers in the United States were the first to determine software system, however, the procurement process was halted due whether machine learning techniques could be used to predict to lack of funds [21]. The judiciary will have to overcome this short courts’ decisions or the voting behavior of judges [59, 60]. Katz term funding obstacle to enjoy the long term cost savings from et al. developed a prediction model that aims to predict whether the utilization of the AI transcription platforms. US Supreme Court as a whole affirms or reverses the status quo AI transcription service systems also provide high searchable judgement, and whether each individual Justice of the Supreme features, allowing for targeted data to be easily identified using Court will vote to affirm or reverse the status quo judgement; the relevant keywords and dispersed files to be consolidated in the form model achieved an accuracy of 70.2% at the case outcome level and of an organized digital database [51]. This would further streamline 71.9% at the justice vote level [61]. Medvedeva et. al. found that case management flow in the ELC and mitigate case backlog. Natural Language Processing techniques could predict (future) judicial decisions in the European Court of Human Rights with an 3.3 Predictive Analysis on Case Duration and Dismissals average accuracy of 75% [62]. The study used a computer to One of the most frustrating aspects of the prolonged adjudication perform quantitative analysis on words and phrases used in a court process in the ELC is that it is just as likely to result in a case case and then based on that analysis trained the computer to predict dismissal as it is to result in a ruling in favor or against the the decision of the Court [62]. It is feasible that similar approaches aggrieved party. In fact, in the survey of ELC done in this study, may be used to predict whether a case may or may not be dismissed the cases that were ultimately dismissed were in the courts longer by the ELC. This approach is likely to reduce the number of land than those that resulted in a judgement for or against one of the dispute cases filed in the ELC – a prediction of dismissal may force parties. the aggrieved parties to seek alternate dispute resolutions. The ethical ramifications of this type of predictive analysis would need to be taken into account, however. There are concerns that the use of this sophisticated AI prediction models may only be accessible to wealthy litigants, leaving those that cannot afford them in a less powerful position of legal armament [63]. Issues of AI bias are well documented [64]; pro-active measures will need to be implemented to identify any bias present in the predictive AI platforms. Predictive analytics may also be used to predict the duration of a court case [65]. This would allow courts to give priority to cases that are predicted to consume less processing time in order to reduce the average total time in adjudicating cases on the docket. Knowledge on the duration of a case might also quell litigants’ desire to submit cases to the ELC, instead seeking alternate routes of dispute resolution. 3.4 Online Dispute Resolution There are 3 reasons that the Kenyan judiciary should embrace online courts as platforms for providing justice. First, some of the farthest regions of the country, largely rural areas, still do not have Figure 5: The average age of cases in ELC in various counties physical court buildings, which means that advocates and witnesses in 2 categories: cases that are ultimately dismissed and those travel long distances in search of justice [66]. In fact, Environment that are heard by the court. In general, cases that were and Land Courts are only present in 29 counties in the country. Second, in instances where legal representation is cost prohibitive, Utilizing AI to Improve Efficiency of the Environment and Land LegalAIIA’21, June, 2021, Sao Paolo, Brazil Court in the Kenyan Judiciary litigants may use these online courts and represent themselves. new platforms of justice introduced by the judiciary. In the coming Finally, online courts may expedite adjudication of a case by years, with greater exposure, familiarity and uptake, ODR may eliminating the need of legal counsel or judges for hearings by fully become the commonly chosen, if not the default option, of automating the legal process or requiring human input only in the arbitration in the ELC. This is especially likely given the sluggish ‘ruling’ portion of the proceedings. In this case, Online Dispute pace of construction of ELC courts across the country and the Resolution (ODR) utilizing online courts would provide means of staggering distances that parties in more remote locales must travel settling land disputes via a hearing using technology but outside of for in – person hearings. the courtroom. There are several such AI platforms in use. ODR platforms such as Rechtwijzer, MyLaw BC, and the British 5 CONCLUSION Columbia Civil Resolution Tribunal, utilize AI to determine which Kenya’s judiciary is stuck in a perennial battle against an ever cases may be adjudicated using the platform, and to automate increasing case backlog. Despite small successes in recent years in decision-making and settlement or outcome proposal [67]. Similar reducing the number of cases pending in the court system, it is clear platforms could be deployed by the Kenyan judiciary system. By that human efforts alone are not sufficient to tackle this problem. mining data from prior related court cases and decisions these AI offers a great opportunity for the judiciary to achieve its service platforms could autonomously decide settlement options or fair delivery goals. The problems caused by insufficient funding and workforce could be mitigated by utilizing AI tools. adjudication. Judges could review the platform’s decision to ensure In this paper we introduce 4 ways that AI may be used to ease it is just. AI could help parties reach an equitable settlement in land the pressure on the Environment and Land Courts; emphasizing disputes. that augmenting these tools to existing human abilities would be The issue of digital literacy, especially in rural counties, must the best way of leveraging both AI and human abilities. While the be considered and addressed for the government to effectively country’s poor digital infrastructure and data capacity does create utilize ODR platforms across the country. The number of people in obstacles in deploying these AI tools, we believe that these are not the country who are able to effectively use digital technologies is insurmountable and that the strategies outlined in this paper are the still low; only 25% of the population are mobile internet users best way forward. according to a 2019 study compared to 95% of the population in the USA in the same year [68].The absence of networks in many REFERENCES rural counties means that fewer people acquire devices such as [1] C. Leonardi and A. Browne, "Valuing Land in Eastern computers or laptops which in turn feeds into the high rates of Africa," Critical African Studies, vol. 10, no. 1, pp. 1 - 13, digital illiteracy in these communities. The ELC may have to tailor 2018. its approach in deploying AI online resolution tools in courts in [2] Land Development and Governance Institute, "An rural counties. The interface for these applications must be simple Assessment of the Performace of the Environment and Land enough to use so that even people with limited skills will find them Court," Land Development and Governance Institute, easy to navigate. Moreover, in deference to the poor digital Nairobi, Kenya, 2014. infrastructure in rural counties, the ODR platforms deployed should work on simple smart phones or other devices that can work with [3] P. M. Syagga, "Land ownership and use in Kenya: Policy lower bandwidth Wi-Fi and don’t need constant access to the power prescriptions from an inequality perspective.," Readings on grid. It is also crucial that the systems be user friendly to self- Inequality in Kenya. Sectoral Dynamics and Perspectives, represented litigants as well as those represented by law firms. Nairobi., pp. 289 - 344, 2006. Finally, there is evidence that ODR platforms are likely to be [4] T. Ndege, "Evolving land tenure and agricultural systems.," embraced by the Kenyan populace. With the onset of the COVID in Ethnography of the Gusii of western Kenya: A Vanishing – 19 pandemic many courts in Kenya are engaging in some form of cultural heritage., Lewistion, NY, Edwin Mellen Press. , ODR, from communicating with litigants via email, to utilizing 2006. electronic disclosure platforms to manage disclosure, submission [5] R. Southall, "The Ndungu report: land & graft in Kenya.," of documents via online portals, and even providing rulings online. Review of African Political Economy, vol. 32, no. 103, pp. These new procedures have ultimately been accepted, begrudgingly 142 - 151, 2005. in some instances, within the legal sphere and by the general public. [6] C. Greiner, "Pastoralism and Land‐Tenure Change in Given this acceptance of various ODR practices by both the public Kenya: The Failure of Customary Institutions.," and legal personnel, it is reasonable to conclude that online conflict Development and Change., vol. 48, no. 1, pp. 78 - 97, 2017. resolution platforms would be similarly embraced. A 2020 study also found that a majority of Kenyans are satisfied with justice [7] J. Munei and K. Galaty, "Maasai Land, Law, and outcomes from the various avenues from where they seek justice – Dispossession," Cultural Survival Quarterly Magazine, this does not have to be in court or even within the conventional December 1998. [Online]. Available: court system [69]. In fact, in general, the citizenry has greater trust https://www.culturalsurvival.org/publications/cultural- in the integrity of the judiciary compared to other governmental survival-quarterly/maasai-land-law-and-dispossession. institutions and because of this may be more willing to embrace [Accessed 21 April 2021]. LegalAIIA’21, June, 2021, Sao Paolo, Brazil F. Ogonjo et al. [8] Ministry of Lands, "“Sessional Paper No. 3 of 2009 on [23] C. Oyugi, "Kenyan Courts Record 84 % Reduction in Case National Land Policy," Ministry of Lands, Nairobi, Kenya, Backlog," Reporting on Good Governance in Kenya., 21 2009. August 2019. [Online]. Available: [9] T. Yamano and K. Deininger, "Land Conflicts in Kenya: https://roggkenya.org/2019/08/21/9720-2/. [Accessed 30 Causes, Impacts, and Resolutions.," Foundation for April 2021]. Advanced Studies on International Development, 2005. [24] C. J. D. Maraga, Interviewee, CJ Maraga: I have no [10] J. Mwita, “Ethnic Land Conflict a Constant Struggle in problem with President Uhuru. [Interview]. 23 August Kenya: A Critical inquest on the role played by the 2020. Methodist church in Meru County, Kenya., Norwegian [25] N. Kihiu, "Land, Environment Courts In Nairobi Booked To School of Theology, 2017. March 2021 Amid Acute Staffing Gap.," Capital News, 23 [11] T. Janus, "Natural Resource Extraction and Civil Conflict.," January 2020. [Online]. Available: Journal of Development Economics, vol. 97, no. 1, pp. 24 - https://www.capitalfm.co.ke/news/2020/01/land- 31, 2012. environment-courts-in-nairobi-booked-to-march-2021- amid-acute-staffing-gap/. [Accessed 27 April 2021]. [12] K. Kanyinga, "Politics and Struggles for Access to Land: “Grants from Above” and “Squatters” in Coastal Kenya.," [26] A. Adeyoju, "Artificial Intelligence and the Future of Law Kanyinga, K. 1998. “Politics and Struggles for Access to Practice in Africa," 15 December 2018. [Online]. Available: Land: “GranEuropean Journal of Development Research , http://dx.doi.org/10.2139/ssrn.3301937. [Accessed 26 April vol. 10, pp. 50 - 70, 1998. 2021]. [13] C. Greiner, "Guns, Land, and Votes: Cattle Rustling and the [27] LawPavillion, "LawPavilion Launches the First Legal Politics of Boundary (Re)making in Northern Kenya," Analytics Software in Africa at the 56th Annual General African Affairs, vol. 112, no. 447, p. 216–237, 2013. Conference of the NBA In Port-Harcourt.," LawPavillion, 2016. [Online]. Available: [14] J. Klopp, "Pilfering the Public: The Problem of Land http://lawpavilion.com/blog/lawpavilionlaunches-the-first- Grabbing in Contemporary Kenya," Africa Today, vol. 47, legal-analytics-software-in-africa-at-the-56th-annual- no. 1, pp. 7 - 26, 2000. general-conference-of-the-nba-in-portharcourt/. [Accessed [15] W. Abuya, "Resource Conflict in Kenya’s Titanium Mining 26 April 2021]. Industry: Ethno-ecology and the Redefinition of [28] P. Ifeoma, "LawPavilion Unveils ‘TIMI’; Nigeria’s First Ownership, Control, and Compensation.," Development Artificial Intelligence Legal Assistant.," DNL Legal and Southern Africa., vol. 34, no. 5, pp. 593 - 606, 2017. Style, 21 August 2018. [Online]. Available: [16] P. A. Nyong'o, "The land question, land grabbing and https://dnllegalandstyle.com/2018/lawpavilion-unveils- agriculture in Africa.," International Journal of African timi-nigerias-first-artificial-intelligence-legal-assistant/. Renaissance Studies-Multi-, Inter-and Transdisciplinarity., [Accessed 27 April 2021]. vol. 8, no. 2, pp. 23 - 31, 2013. [29] Business Daily, "‘Land Ministry in Fresh Bid to Digitise [17] A. Manji, "The grabbed state: lawyers, politics and public Records.," Business Daily, 18 March 2018. [Online]. land in Kenya.," The Journal of Modern African Studies., Available: pp. 467 - 492, 2012. https://www.businessdailyafrica.com/economy/Land- [18] L. Apiyo, "Land-grabbing” and evictions in Kenya.," ministry-in-fresh-bid-to-digitise-records/3946234- Habitat Debate., vol. 4, no. 1, pp. 18 - 19, 1998. 4347056-goa3sg/index.html. [Accessed 26 April 2021]. [19] "Environment and Land Courts," Judiciary, [Online]. [30] "Lands Ministry Wants Orders Stopping Digitisation Available: Lifted," Daily Nation, 2018. [Online]. Available: https://www.judiciary.go.ke/courts/environment-and-land- https://www.nation.co.kehttps/www.nation.co.ke/dailynati court/. [Accessed 24 April 2021]. on/news/lands-ministry-wants-orders-stopping- [20] The Judiciary of Kenya, "State of the Judiciary and the digitisation-lifted--38664. [Accessed 26 April 2021]. Administration of Justice Annual Report 2018 - 2019," The [31] K. Muthoni, "Lawyers sue ministry over digitisation of land Judiciary of Kenya, Nairobi, Kenya, 2019. records," The Standard, 17 April 2018. [Online]. Available: [21] The Judiciary of Kenya, "State of the Judiciary and the https://www.standardmedia.co.ke/nairobi/article/20012772 Administration of Justice Annual Report 2019 - 2020," The 26/lsk-sues-ministry-over-online-land-transactions. Judiciary of Kenya, Nairobi, Kenya, 2020. [Accessed 26 April 2021]. [22] J. Makau, Factors Influencing Managemet of Case Backlog [32] K. Mwanza, "Kenyan lawyers wrangle with government in Judiciary in Kenya: A Case of Courts within Meru and over land registry digitization," Thomson Reuters Tharaka Nithi Counties., Nairobi, Kenya: University of Foundation, 7 May 2018. [Online]. Available: Nairobi, 2014. https://www.reuters.com/article/us-kenya-landrights- idUSKBN1I81K1. [Accessed 30 April 2021]. Utilizing AI to Improve Efficiency of the Environment and Land LegalAIIA’21, June, 2021, Sao Paolo, Brazil Court in the Kenyan Judiciary [33] K. Muigua, "Legal Practice and New Frontiers: Embracing [44] Ministry of Information, Communications and Technology, Technology for Enhanced Efficiency and Access to "National Information, Communications and Technology Justice," June 2020. [Online]. Available: (ICT) Policy," Ministry of Information, Communications http://kmco.co.ke/wp-content/uploads/2020/06/Legal- and Technology, Nairobi, Kenya, 2019. Practice-and-New-Frontiers-Embracing-Technology-for- [45] J. Mugendi, "Challenges in Implementing Digital Enhanced-Efficiency-and-Access-to-Justice-Kariuki- Technologies in Rural Kenya," Engineering for Change, 15 Muigua-Ph.D-June-2020.pdf. [Accessed 26 April 2021]. December 2020. [Online]. Available: [34] D. Lat, "How Artificial Intelligence is Transforming Legal https://www.engineeringforchange.org/news/challenges- Research," Above the Law 2020, [Online]. Available: implementing-digital-technologies-rural-kenya/. [Accessed https://abovethelaw.com/law2020/how-artificial- 10 June 2021]. intelligence-is-transforming-legal-research/. [Accessed 30 [46] D. Louisell and M. Pirsig, "The Significance of Verbatim April 2021]. Recording of Proceedings in American Adjudication," [35] M. Neary and S. Chen, "Artificial Intelligence: Legal Minnesota Law Review, vol. 1100, 1953. Research and Law Librarians," AALL Spectrum, vol. 21, no. [47] Star Reporter, "State to digitise courts, create jobs for 5, pp. 16 - 20, 2017. youth," The Star, 15 March 2019. [Online]. Available: [36] A. Wilkinson, "How Natural Language Processing Can https://www.the-star.co.ke/news/2019-03-15-state-to- Improve Legal Search Results," Kira, 20 July 2020. digitise-courts-create-jobs-for-youth/. [Accessed 26 April [Online]. Available: https://kirasystems.com/learn/how- 2021]. natural-language-processing-improving-can-improve- [48] B. Worthy, "Word Error Rate Mechanism, ASR legal-search-results/. [Accessed 30 April 2021]. Transcription and Challenges in Accuracy Measurement," [37] N. Black, "Lawyers have a bevy of advanced and AI- GMR Transcription, [Online]. Available: enhanced legal research tools at their fingertips," ABA https://www.gmrtranscription.com/blog/word-error-rate- Journal, 22 November 2019. [Online]. Available: mechanism-asr-transcription-and-challenges-in-accuracy- https://www.abajournal.com/web/article/lawyers-have-a- measurement. [Accessed 30 April 2021]. bevy-of-advanced-and-ai-enhanced-legal-research-tools- [49] Verbit Editorial, "Why AI Transcription is a Court at-their-fingertips. [Accessed 30 April 2021]. Reporter’s Secret Weapon," Verbit, 17 February 2019. [38] National Legal Research Group, Inc., "The real impact of [Online]. Available: https://verbit.ai/why-ai-transcription- using artificial intelligence in legal research.," National is-a-court-reporters-secret-weapon/. [Accessed 26 April Legal Research Group, Inc., Charlottesville, VA, 2018. 2021]. [39] J. Waguma, "Kenya’s Judiciary Ups Online Game," Africa [50] K. Wiggers, "Study finds that even the best speech Legal, 8 June 2020. [Online]. Available: recognition systems exhibit bias," Venture Beat, 1 April https://www.africa-legal.com/news-detail/kenya-s- 2021. [Online]. Available: judiciary-ups-online-game. [Accessed 12 June 2021]. https://venturebeat.com/2021/04/01/study-finds-that-even- [40] The Judiciary of Kenya, "Sustaining Judiciary the-best-speech-recognition-systems-exhibit-bias/. Transformation (SJT): A Service Delivery Agenda 2017 - [Accessed 30 April 2021]. 2020," The Judiciary of Kenya, Nairobi, Kenya, 2017. [51] F. Ilma, "How and Why AI-Powered Transcription Services [41] ICT Authority, "Digitisation," ICT Authority, [Online]. are Taking the Legal Industry by Storm," Tarajama, 30 Available: https://icta.go.ke/digitisation/. [Accessed 30 August 2020. [Online]. Available: April 202`]. https://www.tarjama.com/how-and-why-ai-powered- transcription-services-are-taking-the-legal-industry-by- [42] Ministry of Land and Physical Planning, "Digital land storm/. [Accessed 27 April 2021]. transaction platform to curb fraud, says President Kenyatta," Ministry of Land and Physical Planning, [52] S. Brewster, "New York Times Wirecutter," New York Republic of Kenya, 27 April 2021. [Online]. Available: Times, 30 June 2020. [Online]. Available: https://lands.go.ke/digital-land-transaction-platform-to- https://www.nytimes.com/wirecutter/reviews/best- curb-fraud-says-president-kenyatta/. [Accessed 9 June transcription-services/. [Accessed 9 June 2021]. 2021]. [53] Qual Transcribe, "Legal Transcription Services," Qual [43] J. Kariuki, W. Karugu and M. Okode, "Challenges Facing Transcribe, 2020. [Online]. Available: Digitization Projects in Kenya: Case of Implementation of https://www.qualtranscribe.com/deposition-transcription. National Land Information Management System.," [Accessed 14 June 2021]. International Journal of Technology and Systems, vol. 3, [54] L. Baum, The Puzzle of Judicial Behavior, Ann Arbor, no. 1, pp. 23 - 42, 2018. Michigan: University of Michigan Press., 2009. LegalAIIA’21, June, 2021, Sao Paolo, Brazil F. Ogonjo et al. [55] M. Hall and R. Wright, "Systematic Content Analysis of [68] GSMA, "Mobile Internet Connectivity 2019: Sub-Saharan Judicial Opinions.," California Law Review, vol. 96, no. 1, Africa Factsheet," GSMA, 2019. pp. 63 - 122, 2008. [69] AfroBarometer, "Access to justice in Kenya: experience and [56] F. Schauer, "Prediction and Particularity," Boston perception," IDS-University of Nairobi in partnership with University Law Review, vol. 78, no. 3, pp. 773 - 790, 1998. Katiba Institute, Nairobi, Kenya, 2020. [57] X. Nuo, P. Wang, L. Chen, L. Pan, X. Wang and J. Zhao, "Distinguish Confusing Law Articles for Legal Judgment Prediction.," in Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics., 2020. [58] L. Eliot, "Legal Judgment Prediction (LJP) Amid the Advent of Autonomous AI Legal Reasoning," Cornell University, Ithaca, NY, 2020. [59] D. Katz, "Quantitative legal prediction-or-how i learned to stop worrying and start preparing for the data-driven future of the legal services industry.," Emory Law J., vol. 62, p. 909, 2012. [60] P. Wongchaisuwat, D. Klabjan and J. McGinnis, "Predicting litigation likelihood and time to litigation for patents.," in Proceedings of the 16th edition of the international conference on articial intelligence and law., 2017. [61] D. Katz, M. I. Bommarito and J. Blackman, " A general approach for predicting the behavior of the Supreme Court of the United States.," PloS ONE, vol. 12, no. 4, p. e0174698, 2017. [62] M. Medvedeva, M. Vols and M. Wieling, "Using machine learning to predict decisions of the European Court of Human Rights.," Artifical Intelligence and Law, vol. 28, pp. 237- 266, 2019. [63] A. D'Amato, "Legal Realism Explains Nothing," Northwestern School of Law, Evanston, Illinois, 2010. [64] E. Ntoutsi, P. Fafalios, U. Gadiraju, V. Iosifidis, W. Nejdl, M. E. Vidal, S. Ruggieri, F. Turini, S. Papadoulos, E. Krasanakis, I. Komptasiaris, K. Kinder - Kurlanda, C. Wagner, F. Karimi, M. Fernandez, H. Alani, B. Berendt and S. Staab, "Bias in data‐driven artificial intelligence systems—An introductory survey.," Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery., vol. 10, no. 3, 2020. [65] S. Nagel, "Predicting and Reducing Court-Case Time through Simple Logic," North Carolina Law Review, vol. 60, no. 1, 1981. [66] The Judiciary of Kenya, " State of the Judiciary and the Administration of Justice Annual Report, 2017 – 2018," The Judiciary of Kenya, Nairobi, Kenya, 2017. [67] J. Wu, "AI Goes to Court: The Growing Landscape of AI for Access to Justice," Stanford Legal Design Lab, 6 August 2019. [Online]. Available: https://medium.com/legal- design-and-innovation/ai-goes-to-court-the-growing- landscape-of-ai-for-access-to-justice-3f58aca4306f. [Accessed 27 April 2021].