Legal AI is rapidly becoming part of the environment in which lawyers work. That means legal professionals can’t afford to ignore it.
Clients, businesses, government authorities, law enforcement agencies, and other participants in the legal system are increasingly using generative AI. Its influence may already be present in the information lawyers receive, the evidence they assess, and the decisions their clients make.
For legal professionals, a basic understanding of AI is therefore becoming an important part of practising ethically, competently and responsibly.
Technology competence is becoming a professional expectation
The guidance states that barristers should maintain a sufficient level of competence in technology and AI to understand how these tools may affect their work. Crucially, that expectation applies whether or not they choose to use AI themselves.
Similar approaches have emerged or are being considered in Canada, New Zealand, Ireland, and several US states. The principle is straightforward: lawyers are expected to remain informed about developments that may affect legal practice, including the potential benefits and risks of relevant technology.
There’s currently no single, uniform rule governing technology competence across the Middle East and North Africa. However, the direction of travel is becoming clear.
Courts and arbitration bodies in the region are already setting expectations for the responsible use of AI. The Dubai International Financial Centre Courts’ Practical Guidance Note No. 2 of 2023, for example, addresses several of the issues lawyers need to consider when using AI.
These include:
The message is clear. Using AI doesn’t transfer professional responsibility to the technology. The lawyer remains accountable.
AI literacy doesn’t require technical expertise
The good news is that lawyers don’t need to become data scientists or software engineers.
A practical understanding of a few core concepts can help legal professionals ask better questions, evaluate legal AI tools more critically and identify potential risks before those risks affect clients or legal work.
Here are three terms every legal professional should understand.
Human in the loop
In legal practice, this means a lawyer must review, test and validate an AI-generated response before relying on it or incorporating it into legal work.
AI can support research, summarisation, analysis and drafting, but it can’t take responsibility for the accuracy, relevance or legal soundness of the final output. That responsibility remains with the legal professional.
Human oversight should go beyond a quick read. It may require checking the legal authorities relied upon, confirming that the law is current, testing the reasoning, and ensuring the output is appropriate for the relevant jurisdiction and matter.
Guardrails
Guardrails are safeguards built into an AI system to reduce the likelihood of inaccurate, harmful, or inappropriate outputs. Different tools use different kinds of guardrails. Some may filter unsafe content. Others may limit the information an AI system can access or control how data is processed, stored, and used.
For legal professionals, the presence of guardrails is important, but it shouldn’t be treated as a guarantee.
Lawyers should still understand what safeguards are in place, what information the tool relies on, and where its limitations lie. They should also consider whether the tool is appropriate for confidential, privileged or jurisdiction-specific legal work.
Retrieval-augmented generation
Retrieval augmented generation, commonly known as RAG, is an approach that allows an AI system to retrieve information from a defined collection of sources before producing a response.
This can help ground the AI’s answer in identifiable legal materials rather than relying entirely on the information contained in the underlying language model.
In a legal context, RAG can allow users to trace an answer back to the legislation, judgment, commentary or other legal source supporting it. This makes it easier for the lawyer to verify the response and apply professional judgement.
Quality of the answer still depends on the quality, authority, and relevance of the material the system can retrieve. A citation is useful only when the underlying source is current, authentic, and appropriate for the jurisdiction in question.
Understanding AI is part of managing legal risk
Legal AI literacy isn’t simply about learning how to use a new tool. It’s about understanding how technology may affect confidentiality, accuracy, evidence, professional judgement and client service.
As AI becomes more common across the legal and business landscape, lawyers will increasingly need to recognise when it has influenced the information before them. They’ll also need to understand when an AI-generated output can be trusted, when it must be challenged and when it shouldn’t be used at all.
A basic knowledge of concepts such as human oversight, guardrails and retrieval-augmented generation provides a strong starting point. The ethical use of legal AI begins with informed legal professionals who understand both its potential and its limitations.