Earlier this year, the Academy of Experts, one of the UK’s leading professional bodies for expert witnesses, published Guidance for Expert Witnesses on the use of Artificial Intelligence (AI), reflecting the growing acceptance of AI tools throughout the legal system.
Known in Scotland as 'skilled witnesses', those experts with demonstrable expertise in a body of knowledge can be instructed by parties to provide independent factual and opinion evidence to assist the court in making a decision in a dispute. Given the significance of that role, any transformation that affects how expert evidence is prepared carries serious implications for parties and their solicitors.
We have therefore looked at what the guidance has to say about the interaction of AI and expert evidence from a practical perspective. What is clear is that the Academy translates pre-existing duties into clear expectations and consolidates the now widely accepted position that AI may assist expert work but cannot replace expert judgment.
Established Duties of a Skilled Witness
Underpinning the guidance are the long-established duties of a skilled witness to act independently, objectively and transparently. These principles were outlined by the English courts in The Ikarian Reefer case in 1993 and confirmed as applying in all Scottish civil actions by the UK Supreme Court in Kennedy v Cordia (Services) LLP in 2016.
Importantly, the guidance makes clear that the use of AI does not dilute such responsibilities whatsoever. An expert cannot outsource their judgment or reasoning to the tool that they have chosen to adopt, and any AI-generated output relied upon in the expert’s work ultimately remains their own responsibility.
Duty to Understand
A central message in the guidance is therefore that experts are expected to understand when and how they are using AI. It is suggested that the expert will need to adopt a proactive mindset and consult explanatory information about the technology in order to understand the tools they are using from both a functional and legal or regulatory perspective.
This is consistent with their duties, as experts need to be prepared to explain and justify their use of AI if asked to do so by solicitors and/or the court. In particular, a skilled witness is expected to have knowledge about:
- what the tool is designed to do;
- how it generates outputs;
- what data it processes; and
- what legal or regulatory implications arise from its use.
Diligence from the outset of an instruction is therefore critical. For instance, if planning to use AI to summarise a document, do the terms of use allow its data to be repurposed for system development? The guidance raises this particular question as a primary concern, given the AI implications for confidentiality and privilege, explored in our recent article AI & Legal Professional Privilege: Practical Tips for Use.
Proportionality and Risk
The Academy is also clear that consideration must be given as to what safeguards (if any) need to be implemented to protect against inadvertent breaches of duty. In this regard, the guidance adopts a risk-based framework to address the range of use cases that a skilled witness may adopt AI to undertake. Experts are encouraged to identify the specific purpose for which AI is proposed to be used and to assess the risk level accordingly.
Instances of 'low risk' use include transcription of meetings and document organisation. These are regarded as safe as they are unlikely to have a bearing on the objectivity of the expert's opinion. Whereas, at the 'extreme risk' end, using AI to draft substantive analysis is largely discouraged due to the risk that it usurps the independent reasoning of the skilled witness.
This analysis then allows the expert to consider any safeguards proportionate to the risk posed by a tool. An illustrative example is using AI to conduct data analysis. The extent of verification and human oversight will vary depending on how material the analysis is to the expert's conclusions. The greater the influence on the report, the greater the risk, therefore necessitating more stringent protections.
However, some use cases can never be justified. Some are incompatible with the expert’s role by their very nature, such as violating particular laws and regulations (such as breaching GDPR) or outsourcing the wholesale drafting of the report itself to AI. These are marked as 'prohibited' in the guidance which, at the very least, seeks to preserve the reputation of the professional body as a whole.
It is therefore the expert's responsibility to continuously monitor their approach to comply with their duties and protect the rights of others. Any risk-based assessment should be recorded and explainable throughout the litigation.
Working with Instructing Solicitors
The guidance also has practical implications for the relationship between skilled witnesses and their instructing solicitors, emphasising the impact of AI adoption on an expert's ability to work effectively with those instructing them.
For high risk uses of AI, the guidance anticipates early and ongoing dialogue, including disclosure of intended AI use and discussion of its appropriateness in the context of the case. Even for low risk uses, experts are expected to be prepared to explain and justify their approach if questioned.
This reinforces the importance of:
- clearly defined letters of engagement;
- proactive conversations about methodology; and
- continuous monitoring as the litigation evolves.
Interestingly, Lord Neuberger's foreword to the guidance observes that some may even be criticised for not using AI tools in the future. This consideration of 'too little AI' will also have to be borne in mind by those who are instructing experts due to potential efficiency gains and resulting expenses (costs) implications for clients.
MFMac has a proven track record of working with skilled witnesses during the course of a dispute or litigation, helping to ensure that expert evidence is prepared in a manner that is robust and compliant, while remaining fit for purpose in an evolving technological landscape.
This article has been co-authored by Josh Chambers, trainee solicitor at MFMac.
