AI and the Role of the Court Expert

1. Introduction and main argument

Artificial intelligence (AI) has established a visible presence in both legal and forensic practice. Court experts are increasingly confronted with AI in various ways: as a tool for drafting and structuring reports, as an analytical technique within expert examinations, as a source of new forms of manipulation, or even as an object of investigation in its own right. Yet, in discussions about the implications of these developments for the Netherlands Register of Court Experts (NRGD), AI is often treated as a homogeneous phenomenon.

This article is based on the central argument that the impact of AI on court expertise is not uniform, but depends on the role AI plays within the expert examination process. For that reason, AI does not call for a generic revision of the NRGD framework. Instead, it requires a differentiated approach, whereby it is assessed for each specific role whether general AI literacy is sufficient, or whether additional expertise, specialisation, or even a new field of expertise is required.

By analysing AI along these lines, the NRGD can respond in a proportionate and targeted manner, while maintaining its core values of reliability, verifiability, and independence.

2. AI as a general support tool

2.1. Characteristics of use

In its first role, AI is used as a general support tool. This includes applications such as text editing, summarising case files, translation, or searching large volumes of information. Such use does not directly affect the substantive analysis or the answering of the research questions.

2.2. Significance for the expert

In this role, the expert remains fully responsible for the content of the report and for all professional judgments made. AI functions no differently from other supporting technologies: it enhances efficiency but does not determine the substance of the expert’s conclusions. Nevertheless, this type of use requires a certain level of AI literacy, particularly with regard to the technology’s limitations, potential errors, and the need for critical source evaluation.

2.3. Implications for the NRGD

For the NRGD, this type of use is likely to have limited consequences in the medium term. There is no reason to establish separate fields of expertise or additional registration requirements. AI literacy may be regarded as part of the general professional standard and could, where appropriate, be safeguarded through continuing professional development or professional conduct rules.

The recent addendum on the Use of AI to the NRGD Code of Conduct aligns primarily with this role. Its emphasis on personal responsibility, transparency regarding the use of AI, and avoiding the uncritical adoption of AI-generated output implicitly assumes that AI serves a supportive function. Notably, however, this underlying assumption is not made explicit. This may create confusion, as certain requirements set out in the addendum are not relevant to other AI roles (e.g., “Provide prompts and input upon request”), while requirements that would be relevant to those roles are absent (e.g., validation of AI models).

3. AI as an analytical tool within expert examination

3.1. Characteristics of use

In its second role, AI functions as an analytical tool that makes a substantive contribution to the findings of the expert examination. Examples include image and audio analysis, pattern recognition, and forensic data analysis. In this context, the outputs generated by the AI system directly influence the expert’s conclusions.

his role is comparable to the use of statistical software in DNA analysis: the algorithm performs complex calculations, while the expert remains responsible for the interpretation and justification of the results.

3.2. Significance for the expert

The use of AI as an analytical tool places increased demands on the expert. The expert must have an understanding of the operation, assumptions, and limitations of the system used, as well as of the data on which it is based. They must be able to assess whether the tool is suitable for the specific research question, validate both the system and its application, and interpret and report the results generated by the system.

In some cases, this may require integrating the AI-generated output with the results of other methods, such as human interpretation. In this role, the use of AI directly affects the core of the expert’s professional competence.

3.3. Implications for the NRGD

For the NRGD, this means that AI-intensive specialisations may emerge within existing fields of expertise. This may warrant a refinement of registration and re-registration requirements, as well as assessment criteria that explicitly address transparency, validation, and a thorough understanding of the AI tools employed.

For a more detailed discussion of this role, see: H. van den Heuvel, R.J.F. Ypma, Z.J.M.H. Geradts & J. Meeuwissen, ‘De invloed van AI op forensisch bewijs in strafzaken: kansen en bedreigingen’, EeR 2025/25, p. 127.

4. AI and authenticity examination: deepfakes and detection

4.1. Characteristics of use

The emergence of deepfakes and other AI-generated manipulations has significant implications for authenticity examinations. Traditional methods of image, audio, and document analysis are becoming increasingly insufficient without additional knowledge of AI technologies.

4.2. Significance for the expert

In this domain, AI plays a dual role: as a technology that undermines authenticity and as a tool for its detection. Consequently, all authenticity experts must possess knowledge of the generative AI technologies used to create deepfakes and of the characteristics of such manipulated material.

In addition, specialised expertise in the use of AI-based detection tools is likely to become increasingly valuable, subject to the requirements outlined in the previous section. Given the rapid pace of developments in this field, these experts will need to devote more time to maintaining and updating their knowledge and skills.

4.3. Implications for the NRGD

For this group of experts, AI-related knowledge becomes part of their core competence. This does not constitute a new field of expertise, but rather requires a recalibration of existing expert profiles. The NRGD will need to ensure that registration requirements remain aligned with the current state of the art and that the reporting of uncertainty and probabilistic conclusions is adequately standardised.

For a more detailed discussion of this role, see: H. van den Heuvel, R.J.F. Ypma, Z.J.M.H. Geradts & J. Meeuwissen, ‘De invloed van AI op forensisch bewijs in strafzaken: kansen en bedreigingen’, EeR 2025/25, p. 127.

5. AI as the object of expert examination

5.1. Characteristics of use

In the fourth role, AI itself becomes the subject of expert examination. The issue is not the expert’s use of AI as a tool, but situations in which an AI system itself forms part of the factual circumstances of a case.

Examples include:

• A criminal case in which the police have used a predictive model for risk assessment or hotspot analysis, and the defence challenges the model’s validity or alleges discriminatory effects; 
• An administrative law case in which an automated risk-profiling system (for example, for fraud detection or regulatory supervision) played a significant role in the decision-making process; 
• A civil case in which a credit-scoring or insurance model influenced an individual assessment, raising questions about whether the algorithm used is sufficiently transparent and explainable; 
• A criminal case involving the use of a facial recognition system, where questions arise regarding error rates, training data, and bias. 

In such cases, the central question is not what the system produced in the individual case, but how the system functions:

• What data were used for training? 
• Which assumptions and parameters were built into the system? 
• How was the system validated? 
• What are its known error rates? 
• To what extent is the system susceptible to bias or context dependency? 

The legal relevance lies in the evidential value of the system’s output and the extent to which a decision can be based on that output, given requirements of transparency, explainability, and verifiability.

5.2. A new form of expertise

This role requires experts who combine technical knowledge of AI and data science with normative and contextual analysis. They must be capable of explaining the assumptions, biases, and limitations of AI systems in a manner that is understandable and meaningful to the court.

This form of expertise requires more than purely technical knowledge; it demands an understanding of the context in which the system operates. Unlike the use of AI as an analytical tool (section 3), responsibility here does not primarily rest with the expert using the system, but with the organisation that deployed it. The expert is asked to analyse the system from an external perspective and provide a legal and technical assessment of its operation.

5.3. Implications for the NRGD

This is the area in which a genuine need may arise for a new field of expertise, such as AI and Data Science. Such a development raises questions regarding its delineation from existing fields of digital and statistical expertise, as well as how competence can be assessed in a rapidly evolving domain.

For a more detailed description, see: J. Henseler, ‘De opkomst van kunstmatige intelligentie in expertise en recht: de ‘kras-sporen’ van deeplearning’, EeR 2022, afl. 2, p. 31‑34.

6. Conclusion and recommendations for the NRGD

This article demonstrates that AI does not have a uniform impact on court expertise. A role-differentiated approach clarifies where general AI literacy is sufficient and where additional regulation and guidance are required.

Based on this analysis, the following recommendations can be made:

1. Make role distinctions explicit in policies and guidelines, so that it is clear which forms of AI use fall under which normative frameworks. 
2. Embed general AI literacy through continuing professional development, without elevating it to a registration requirement for all experts. 
3. Develop AI-related specialisations within existing fields of expertise where AI forms a structural component of the methodology. 
4. Reassess expert profiles for authenticity examinations, with explicit attention to deepfakes and AI-based detection tools. 
5. Explore and develop a new field of expertise in Forensic AI and Data Science, with clearly defined boundaries and assessment criteria. 

By adopting this differentiated approach, the NRGD can maintain its role as guardian of reliable, verifiable, and independent expertise, while simultaneously creating space for technological innovation within expert examination.

By Rolf Ypma* in EeR 2026/18

Prof. Dr Ir R. Ypma is Principal Scientist at the Netherlands Forensic Institute (NFI) and Professor of Forensic Data Science at Delft University of Technology (TU Delft).