The Law Foundation of Ontario has granted $467,000 to the Conflict Analytics Lab, led by Professor Samuel Dahan, and the Harvard Access to Justice Lab for a collaborative project to enhance Pro Bono Ontario client access to housing law through a lawyer-controlled AI system.
The Queen’s Conflict Analytics Lab (CAL), in partnership with the Access to Justice (A2J) Lab at Harvard Law School, has received $467,000 through the Law Foundation of Ontario’s Responsive grant program.
Led by CAL Director Professor Samuel Dahan, along with A2J Lab Faculty Director Jim Greiner and Associate Director of Research Innovations Mandi Mobley Li, the project is designed to improve Pro Bono Ontario’s (PBO) client intake and lawyer-client interactions through a lawyer-controlled AI system built on the OpenJustice platform. The tool will help PBO lawyers gather structured information, reduce conversational bottlenecks, and support more efficient housing-law consultations. The Harvard Access to Justice Lab will lead an evaluation of the pilot, including randomized controlled trials, to assess its impact on service quality and capacity.
Titled “OpenJustice: A Community Platform for Building Accessible and Reliable Legal AI,” the two-year pilot project specifically aims to improve housing-law clients’ access to PBO services by streamlining intake and supporting more consistent, timely assistance. Dahan will work with PBO to co-develop a custom intake tool using OpenJustice, an open-source legal AI platform designed to embed explicit legal reasoning into AI systems and help reduce hallucinations and sycophancy. Created in 2021 and grounded in recent frontier AI research on human–AI reasoning integration, OpenJustice allows lawyers to express their decision-making thinking in plain English. Early benchmarking indicates that this approach can improve stability, accuracy, and transparency compared to standard large language models.
The objective of the PBO pilot is to test this method in real client interactions — particularly in housing matters, where clients often have no legal support and the risk of misleading or overly confident AI output is highest.
“Our project addresses the urgent need to enhance the capacity of pro bono organizations, specifically Pro Bono Ontario, to provide actionable information in high-volume areas,” Dahan explains. “A critical area of focus is housing law, where timely and accurate legal assistance is essential to prevent homelessness, protect tenants' rights, and ensure fair treatment in the rental market. Housing disputes, including tenant protection, eviction processes, and rent disputes, are high-demand areas that overwhelm existing legal aid resources. Without adequate support, individuals and families facing these issues often experience severe consequences, such as losing their homes or enduring prolonged legal battles without proper representation.”
An estimated 78 per cent of Canadians with legal problems — nearly six million people —are unable to afford legal representation, and legal aid organizations are forced to turn away 50 per cent of those who approach them. As a result, countless Canadians are left to navigate the legal system on their own, often with little understanding of the law and their rights.
This initiative will focus on developing a custom system that integrates into PBO’s legal hotline, which receives close to 30,000 calls per year. The system will be integrated with PBO’s speech-to-text functionality, allowing lawyers to receive real-time AI-generated suggestions and information during client interactions such as phone consultations. This AI support will assist lawyers by processing the information provided, helping to identify relevant legal precedents, and generating insights, allowing them to focus on delivering expert legal advice to clients.
The project will assist lawyers by automating routine tasks such as triaging cases, answering common legal questions, and generating initial drafts of legal documents. “This automation will enable PBO lawyers to handle a higher volume of cases, providing timely and accurate legal information to clients,” Dahan says.
The CAL team will collaborate with the Harvard A2J Lab to evaluate the effectiveness of the AI tools developed. Randomized controlled trials and computational benchmarking will measure PBO lawyers’ productivity, assessing whether the tools help them work faster and more efficiently while maintaining the quality of advice provided. The accuracy of OpenJustice outputs compared to human-provided legal advice will also be analyzed, focusing on critical issues like fairness and inclusivity. Incorporating Gender-based Analysis Plus (GBA+) and anti-racist perspectives will help ensure the AI’s performance is equitable across different demographics.
Dahan notes that while AI tools are advancing quickly in the private sector, even specialized legal platforms still show substantial error rates. Recent studies, including research from Stanford’s Human-Centered AI Institute, have found that legal AI products can produce incorrect answers in one out of six queries — or more. For public and non-profit organizations, the challenge is compounded by high costs, limited engineering capacity, and difficulty adapting general-purpose models to sensitive legal workflows.
“By leveraging our open-source legal AI platform, this project offers a critical opportunity for the A2J sector in Ontario to catch up with private advancements, ensuring that legal aid services are equipped with state-of-the-art tools to effectively meet community needs,” he says.
“We expect that these tools will improve both the speed and quality of legal advice provided by enabling real-time AI support during consultations. Additionally, PBO’s ability to handle more cases will increase, thanks to the efficiency gains offered by OpenJustice.”
In the long term, Dahan says he and his team plan to replicate this model with other A2J organizations across Canada, including those serving rural and remote populations who face unique barriers to legal services. “By partnering with A2J organizations across the country, we will foster a community-driven approach to the development and deployment of AI tools, allowing organizations to continuously improve their services and increase their capacity to provide legal assistance to underserved populations.”
Dahan recently received funding from the Social Sciences and Humanities Research Council (SSHRC) to fund his project “Exploring Human-AI Methods for Access to Justice,” and OpenJustice recently received funding from the Natural Sciences and Engineering Research Council (NSERC) and project partners totalling more than $1.8 million. Read more in this Queen’s Law article.
By Tracy Weaver