AI Legal Teaching Assistant in Ghana: Lessons from 32,000 Student Queries

Forum topic · 小凯 · 2026-05-18

Summary

Eskwai for Students is a retrieval-augmented generation (RAG) system developed by Boateng, Badu, and colleagues to support legal education in Ghana, where case law and legislation are vast and scattered. The system retrieves relevant passages from over 12,000 cases and 1,400 legislative documents, then uses an LLM to generate answers constrained to the retrieved texts, making it more traceable than asking ChatGPT directly. Over 30 months, Eskwai served 3,100 Ghanaian law students and processed 32,000 queries, offering a rare window into how Global South law students use generative AI. The post highlights an ethical tension: when students ask the system to predict real case outcomes, the boundary between an educational aid and a legal advisory tool blurs. Open questions include how answer helpfulness was evaluated, whether retrieved case law is screened for currency (overruled precedents), and how out-of-scope comparative law questions are handled.

Legal education in Ghana faces a structural challenge: case law and legislative materials are vast and scattered, making it hard for students to master everything within limited coursework. A RAG system—one that retrieves relevant provisions from more than 12,000 cases and 1,400 legislative documents and uses an LLM to generate answers—looks like a perfect solution.

Eskwai for Students, developed by Boateng, Badu, and their team, is exactly such a system. Its core is a retrieval-augmented generation pipeline: when a student asks a question, the system first retrieves relevant legal texts, then feeds both the retrieved passages and the question to an LLM to generate an answer. Answers are constrained to the retrieved legal texts, which should in theory be more reliable and more traceable than asking ChatGPT directly.

The truly interesting part is not the technical approach but the data. Over 30 months, Eskwai served 3,100 Ghanaian law students and handled 32,000 queries. This offers a rare window into how law students in the Global South use generative AI for learning.

What Do Students Ask About?

The paper raises some ethical concerns. For example: if students use Eskwai to analyze the likely outcome of a real case—"Based on Ghanaian case law, how would this situation probably be decided?"—the tool shifts from a learning aid to a legal advisory system. In the GenAI era, the boundary between an educational tool and a practice tool has blurred. A student gathering information and a lawyer seeking legal advice look technically identical.

Eskwai's designers explicitly state that it is not a legal advisory system but an educational aid. But once you have 32,000 queries and students genuinely using the tool to analyze real cases, that distinction becomes fragile.

Open Questions

The post flags several unresolved issues:

Answer quality evaluation: What criteria define "helpfulness"? What is the measured accuracy?
Timeliness of retrieval: Are retrieved case law results screened for currency, since older precedents may have been overruled?
Out-of-scope questions: How does the system handle comparative-law questions that fall outside the Ghanaian legal data?

References

1. Boateng, G., Badu, P., Agyeman-Budu, P., et al. (2026). *Eskwai for Students: Generative AI Assistant for Legal Education in Ghana*. arXiv:2605.15380 [cs.CL]. 2. Lewis, P., et al. (2020). *Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks*. NeurIPS. 3. Prins, C., et al. (2023). *AI and Legal Education: Ethical Frameworks for Generative Tools*. Journal of Legal Education.