pith. sign in

arxiv: 2605.15380 · v1 · pith:OQ6GKRLYnew · submitted 2026-05-14 · 💻 cs.CL · cs.CY· cs.HC

Eskwai for Students: Generative AI Assistant for Legal Education in Ghana

George Boateng , Philemon Badu , Patrick Agyeman-Budu , Samuel Ansah , Evans Atompoya , Evan Igwilo , Lord Baah , Frederick Abu-Bonsrah
show 1 more author
Victor Wumbor-Apin Kumbol
This is my paper

Pith reviewed 2026-05-19 15:12 UTC · model grok-4.3

classification 💻 cs.CL cs.CYcs.HC
keywords generative AIlegal educationGhanalaw studentscase lawslegislationAI in educationethical concerns
0
0 comments X p. Extension
pith:OQ6GKRLY Add to your LaTeX paper What is a Pith Number? 
\usepackage{pith}
\pithnumber{OQ6GKRLY}

Prints a linked pith:OQ6GKRLY badge after your title and writes the identifier into PDF metadata. Compiles on arXiv with no extra files. Learn more

The pith

A generative AI assistant helps Ghanaian law students by answering questions using a database of local case laws and legislation.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper reports on building and testing an AI tool that supports legal education in Ghana by answering student questions. The system draws from a collection of more than twelve thousand case laws and fourteen hundred pieces of legislation from the country. A study tracked its use by over three thousand students across thirty months, during which they submitted thirty-two thousand queries. The authors looked at how helpful the responses were and what kinds of questions came up, which also brought up some ethical issues around AI in education. If this works as described, it offers a model for using AI to improve access to legal knowledge in regions with fewer traditional resources.

Core claim

The central claim is that a generative AI system, built to retrieve and use Ghanaian legal documents, can assist law students effectively, as evidenced by its sustained use over two and a half years by thousands of students who asked tens of thousands of questions, with evaluations showing its value while highlighting query patterns and ethical considerations.

What carries the argument

The mechanism that searches a curated database of Ghanaian case laws and legislation to find relevant information and then generates answers based on those documents.

If this is right

  • Students receive grounded responses to legal questions without needing immediate access to full court records or statutes.
  • Usage data reveals common topics and difficulties that law students face in their studies.
  • Ethical concerns arise from the types of questions asked, requiring careful guidelines for tool use.
  • The long deployment period demonstrates practical feasibility in a real educational setting in Ghana.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar AI assistants could be developed for legal education in other countries facing resource constraints.
  • Future work might examine whether students who use the tool perform better in exams or understand concepts more deeply.
  • Pairing the AI with traditional teaching methods could help manage risks of over-reliance on the technology.

Load-bearing premise

The database of case laws and legislation remains accurate, complete, and up to date so the AI does not give students wrong or outdated information.

What would settle it

A review by Ghanaian legal experts finding frequent errors or omissions in the AI's answers to standard questions about key cases and laws would undermine the claims of helpfulness.

Figures

Figures reproduced from arXiv: 2605.15380 by Evan Igwilo, Evans Atompoya, Frederick Abu-Bonsrah, George Boateng, Lord Baah, Patrick Agyeman-Budu, Philemon Badu, Samuel Ansah, Victor Wumbor-Apin Kumbol.

Figure 1
Figure 1. Figure 1: Screenshots of Eskwai 3 System Overview Eskwai for Students 4 consists of a web app component and an AI component. 3.1 Web App The web app ( [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Screenshots of Mobile Version of Eskwai a commercial model (the models used have changed over the years). It generates a response to queries, grounded in a database of cases and legislation, as well as user documents, and provides inline citations to the exact passages used to generate the response. We created 5-sentence chunks from a corpus of cases (12K) and legislation (1.4K), which we curated primarily… view at source ↗
Figure 3
Figure 3. Figure 3: Overview of Eskwai RAG System 4.1 Usability During the evaluation period, Eskwai recorded significant engagement from the law student community. A total of 3,127 students across more than 15 Ghanaian universities utilized the platform, collectively submitting 32,919 queries on Ask Kwame. Most users accessed Eskwai on mobile devices. Users could provide upvotes or downvotes on the answers in response to a q… view at source ↗
read the original abstract

Recent advances in generative AI have shown their potential to be leveraged for legal education. Yet, work on the development and deployment of such systems for legal education in the Global South is limited. In this work, we developed Eskwai for Students, a generative AI assistant to help law students with their legal education. Eskwai for Students is a retrieval augmented generation (RAG) system that provides answers to a wide range of legal questions for law students grounded in a curated database of over 12K case laws and 1.4K legislation in Ghana. We deployed Eskwai for Students in a longitudinal study of 30 months (2.5 years) used by 3.1K law students in Ghana who made 32K queries. We evaluated the helpfulness of our AI, and provided insight into the kinds of queries law students submit to this generative AI tool, which raises some ethical concerns. This work contributes to an understanding of how law students in the Global South are using generative AI for their studies and the ways it could be leveraged responsibly to advance legal education.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript describes the development of Eskwai for Students, a retrieval-augmented generation (RAG) system for law students in Ghana. It is grounded in a curated database of over 12K case laws and 1.4K pieces of legislation, and was deployed over 30 months to 3.1K students who submitted 32K queries. The authors report an evaluation of the system's helpfulness, analysis of query types submitted by students, and identification of associated ethical concerns, positioning the work as a contribution to understanding generative AI use in legal education in the Global South.

Significance. If the reported deployment and insights hold, the work provides a useful existence proof and usage study of a large-scale RAG deployment in a Global South legal-education context. The longitudinal scale (30 months, 3.1K users, 32K queries) and focus on query patterns plus ethical observations add practical value to the limited literature on such systems outside high-resource settings. The absence of detailed evaluation protocols, however, limits the strength of the performance and insight claims.

major comments (1)
  1. [Abstract] Abstract (and corresponding evaluation section): the manuscript states that helpfulness was evaluated and ethical concerns were identified, yet supplies no description of the evaluation method, metrics, sample size, query sampling procedure, or how ethical issues were defined and measured. This omission is load-bearing for the central claims about system performance and the insights derived from usage data.
minor comments (2)
  1. [System Description] Clarify the curation process, update frequency, and coverage gaps of the 12K case-law / 1.4K legislation database so readers can assess its fitness for grounding student queries.
  2. [Deployment and Usage] Provide basic descriptive statistics (e.g., query distribution by topic, response latency, or user retention) to support the longitudinal-deployment narrative.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful review and for identifying a key area where the manuscript can be strengthened. We agree that the evaluation methods and protocols require explicit description to support the claims regarding helpfulness and ethical insights. We will revise the manuscript to address this directly.

read point-by-point responses

  1. Referee: [Abstract] Abstract (and corresponding evaluation section): the manuscript states that helpfulness was evaluated and ethical concerns were identified, yet supplies no description of the evaluation method, metrics, sample size, query sampling procedure, or how ethical issues were defined and measured. This omission is load-bearing for the central claims about system performance and the insights derived from usage data.

    Authors: We acknowledge that the current version of the manuscript does not provide sufficient detail on the evaluation process in either the abstract or the corresponding section. This is a valid observation that weakens the transparency of our performance and insight claims. In the revised manuscript, we will expand the evaluation section (and update the abstract accordingly) to include: a clear description of the evaluation method (human assessment by domain experts combined with automated checks), the specific metrics employed (e.g., Likert-scale ratings for relevance, accuracy, and completeness), the sample size and selection criteria for the evaluated queries, the sampling procedure used to select queries from the full set of 32K, and the framework for identifying and categorizing ethical concerns (via thematic coding of query patterns). These additions will be placed in a new subsection to ensure the central claims are properly grounded. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper is a descriptive report of system construction, deployment of a RAG-based legal AI assistant, and analysis of 32K student queries over 30 months. It contains no equations, derivations, fitted parameters, predictions, or uniqueness theorems. The central claims rest on the existence of the curated database and observed usage data rather than any self-referential reduction or self-citation chain that would force the reported outcomes by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper introduces no free parameters or invented entities. It relies on the standard assumption that retrieval-augmented generation improves factual grounding when supplied with a domain-specific corpus.

axioms (1)
  • domain assumption A curated database of Ghanaian case laws and legislation can serve as a reliable grounding source for accurate answers to student legal questions via RAG.
    The system's claimed usefulness and the interpretation of query data rest on the quality and completeness of this database, which is stated without further validation in the abstract.

pith-pipeline@v0.9.0 · 5766 in / 1623 out tokens · 81994 ms · 2026-05-19T15:12:35.981442+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

  1. [1]

    The Law Teacher57(3), 352–364 (2023)

    Ajevski, M., Barker, K., Gilbert, A., Hardie, L., Ryan, F.: Chatgpt and the future of legal education and practice. The Law Teacher57(3), 352–364 (2023)

    work page 2023

  2. [2]

    Second Draft37, 1 (2024)

    Alsbrook, M., Chase, A.K.: Three blind drafts: An ai-generated classroom exercise. Second Draft37, 1 (2024)

    work page 2024

  3. [3]

    International Journal of the Legal Profession31(3), 323–348 (2024)

    Balan, A.: Examining the ethical and sustainability challenges of legal education’s ai revolution. International Journal of the Legal Profession31(3), 323–348 (2024)

    work page 2024

  4. [4]

    Jurimetrics Journal64, 111–134 (2024)

    Bliss, J.: Teaching law in the age of generative al. Jurimetrics Journal64, 111–134 (2024)

    work page 2024

  5. [5]

    Law, Technology and Humans 6(3), 5–22 (2024)

    Burgess,P.,Williams,I.,Qu,L.,Wang,W.:Usinggenerativeaitoidentifyarguments in judges’ reasons: Accuracy and benefits for students. Law, Technology and Humans 6(3), 5–22 (2024)

    work page 2024

  6. [6]

    arXiv preprint arXiv:2603.04982 (2026)

    Chen, B.M., Bao, H.: Training for technology: Adoption and productive use of generative ai in legal analysis. arXiv preprint arXiv:2603.04982 (2026)

    work page arXiv 2026

  7. [7]

    International Journal of the Legal Profession31(3), 349–363 (2024)

    Farber, S.: Harmonizing ai and human instruction in legal education: a case study from israel on training future legal professionals. International Journal of the Legal Profession31(3), 349–363 (2024)

    work page 2024

  8. [8]

    arXiv preprint arXiv:2508.09713 (2025)

    Hemrajani, R.: Evaluating the role of large language models in legal practice in india. arXiv preprint arXiv:2508.09713 (2025)

    work page arXiv 2025

  9. [9]

    73 Journal of Legal Education (forthcoming (2023)

    Jonathan, H., Daniel, S., et al.: Ai assistance in legal analysis: An empirical study. 73 Journal of Legal Education (forthcoming (2023)

    work page 2023

  10. [10]

    Center for Law & Economics Working Paper Series10(2024)

    Nielsen, A., Skylaki, S., Norkute, M., Stremitzer, A.: Building a better lawyer: experimental evidence that ai can increase legal work efficiency. Center for Law & Economics Working Paper Series10(2024)

    work page 2024

  11. [11]

    Regalia, J.: From briefs to bytes: How generative ai is transforming legal writing and practice. Tulsa L. Rev.59, 193 (2024)

    work page 2024

  12. [12]

    Law, Innovation and Technology pp

    Schrepel, T.: Generative ai in legal education: A two-year experiment with chatgpt. Law, Innovation and Technology pp. 1–42 (2026)

    work page 2026

  13. [13]

    Fordham L

    Surden, H.: Chatgpt, ai large language models, and law. Fordham L. Rev.92, 1941 (2023)

    work page 1941