The Civil Justice Council is consulting on the use of artificial intelligence in the preparation of court documents, I have been reflecting on what it means in practice.
As a forensic accountant, some of the questions the consultation raises feel particularly relevant to the work I do. Loss quantum work involves expert opinion built from experience and analysis of financial evidence and records.
So when we talk about AI in expert evidence, the question for me is not just whether it helped draft the report. It goes to how the analysis was done in the first place.
Some context around AI and Expert Evidence
As we have all seen, AI has become a hot topic in professional work, with new AI tools released all the time, and AI features being added to many commonly used software products. The legal community is now considering how it should be disclosed in court proceedings. The Civil Justice Council’s consultation on the use of artificial intelligence in the preparation of court documents raises important practical questions for those involved in litigation, including expert witnesses.
For those working as financial expert witnesses, many of the issues raised by AI will feel familiar. The work can often involve analysing large volumes of financial information and using a range of analytical tools as part of that process.
There is also a well-established practice within accountancy and forensic work for elements of the analytical process to be undertaken by professional staff working under the supervision of the reporting expert.
In accountancy firms this may involve members of the forensic or accounting team assisting with aspects of the work, such as reviewing financial documents, carrying out investigative tasks, preparing financial analysis, supporting interviews, or undertaking site visits. The expert reviews this work, applies independent professional judgement, and takes responsibility for the conclusions expressed in the report.
In other cases, experts operate as sole practitioners and undertake this work themselves. The structure of the practice may therefore differ, but the underlying principle remains the same: the expert is responsible for the opinion presented to the court.
Seen in that context, the emergence of AI tools represents less a fundamental change than another step in the continuing evolution of the tools used in professional analysis.
AI tools can therefore be viewed as another form of analytical assistance, provided that the expert retains control of the reasoning and conclusions.
The expert’s responsibility must remain the central principle
At the heart of expert evidence is a simple principle that the expert is responsible for the opinion expressed in the report.
That responsibility does not change because new technology is used during preparation.
The expert must remain the author of the report, responsible for the opinions expressed within it, and responsible for verifying the evidence and analysis relied upon.
Any AI output must therefore be reviewed and verified by the expert before it forms part of the reasoning in the report.
Existing duties of expert witnesses already require experts to take responsibility for the evidence relied upon and the reasoning that supports their conclusions. These obligations remain the primary safeguard.
AI is another professional tool
Financial expert work has long involved the use of tools and assistance during report preparation.
Examples include analytical software used to process financial information, working papers and financial schedules prepared as part of the analytical process, and document review and analysis undertaken as part of investigations.
Courts have historically accepted these practices provided that the expert retains control over the analysis and the final conclusions.
In that sense, many AI tools perform functions comparable to analytical or drafting assistance already used in professional practice. The key issue should therefore remain expert supervision and responsibility, rather than the mere use of technology.
The distinction between ‘administrative’ and ‘substantive’ AI may be unclear
Some discussions distinguish between administrative uses of AI, such as transcription or formatting, and more substantive uses affecting the expert’s analysis.
In practice, the boundary between these categories may not always be clear.
Experts may use tools to organise and classify large volumes of documents, assist with spreadsheet modelling or calculations, review draft reports for structure or clarity, identify omissions or inconsistencies, and highlight potential gaps in reasoning.
These functions assist with preparing a report but do not determine the expert’s opinion.
For that reason, it may be more practical to refer more broadly to administrative or supportive uses rather than attempting to draw a rigid line between categories.
Financial experts may already rely heavily on analytical software
Financial expert work frequently involves specialist systems designed to analyse financial records. Tools like accounting or analysis software, like Dext are good examples of this, with AI functionality built in as standard to process invoices, extract financial information and categorise data.
Such tools have long been accepted as part of the analytical process and are not normally disclosed in expert reports.
This points to a broader issue as AI is not new. It has been embedded in professional software for years, long before the recent wave of chatbots and large language models made it a topic of public debate.
But increasingly its use is not optional. AI is now built into email programmes, file storage and document management systems, for example, now Microsoft and Google have embedded it across their products as standard. In many cases experts are not choosing to use AI at all, it is simply there in the tools they already rely on.
As AI functionality becomes further embedded within commonly used software, requiring experts to identify every instance of AI-assisted functionality may quickly become impractical.
AI disclosure should align with existing expert obligations
Experts already operate within a well-established framework under the Civil Procedure Rules.
In particular, experts must identify the sources relied upon and disclose those who have assisted in preparing the report.
Any new requirements relating to AI would therefore sit most naturally within the existing framework governing expert evidence.
The statement of truth confirms the honesty and independence of the expert’s opinion. It was not designed to catalogue the technical tools used during report preparation.
Expert reports traditionally focus on the evidence relied upon and the reasoning supporting the conclusions, rather than the mechanics of the analytical process.
The definition of AI presents practical challenges
A further difficulty is the breadth of the term artificial intelligence.
AI functionality is increasingly embedded within everyday software tools, and users may not always know when such functionality is operating.
Rules requiring disclosure of AI use therefore need to be framed carefully to avoid creating obligations that may be difficult for experts to identify or comply with in practice.
Technology in professional work continually evolves
Professional practice has always evolved alongside technological change.
Experts routinely use electronic document systems, internet research tools and data analytics software. These are not normally disclosed to the court. Instead, experts disclose the sources relied upon and the reasoning behind their conclusions.
In many respects, AI represents a further development of existing research and analytical tools rather than a fundamentally new category of technology.
A proportionate approach to AI disclosure in expert evidence
A practical and proportionate approach may therefore be: no disclosure for routine administrative or analytical assistance, and disclosure only where AI has materially contributed to the substantive analysis or conclusions of the expert report.
Ultimately, the essential safeguard in expert evidence is unchanged: the expert’s independent judgement and responsibility for the opinion presented to the court.
Richard Formby is a forensic accountant and expert witness. This article draws on his response to the Civil Justice Council consultation on the use of artificial intelligence in the courts.
Richard Formby FCA MAE
Get in touch to discuss your client’s case
