Why AI can't substitute human judgment in psycho-legal work - part 1

Image source: Freepik

Recent discourse around the role of artificial intelligence (AI) in psycho-legal reporting has introduced confusion, particularly in contexts governed by legal and professional mandates.

Much of this stems from commentary that omits a crucial fact: psycho-legal work by industrial and organisational psychologists (IOPs) is a regulated psychological act under the jurisdiction of the Health Professions Council of South Africa (HPCSA).

This is not a general exploration of AI in psychology, nor a speculative debate about automation. It is a necessary and discipline-specific response intended to clarify the unique function and scope of the IOP in psycho-legal matters. That function is often misunderstood, and when mischaracterised in published content, the implications can be both professionally misleading and legally consequential.

The role of the IOP in this context is not interchangeable with data analysis, actuarial forecasting, or diagnostic psychology. Nor can it be outsourced to predictive platforms, automated assessments, or AI-generated approximations.

Each psycho-legal opinion rendered by an IOP must be independently reasoned, ethically grounded, and defensible within a legal forum. And while AI may support certain peripheral processes, it cannot assume the central cognitive, professional, and evaluative responsibilities that define this scope of practice.

The scope of the industrial psychologist in psycho-legal matters

In psycho-legal matters, the IOP operates under a defined and regulated mandate, grounded in HPCSA registration and the application of psychological knowledge specific to human functioning within the world of work.

This work is not clinical, actuarial, or administrative. It is forensic by nature, psychological by foundation, and requires integration of verified data with expert judgment to provide an opinion on employability, career trajectory, and earning capacity – pre- and post-incident.

The IOP’s unique scope is governed by ethical rules, professional standards, and a body of case law that recognises the practitioner’s responsibility in producing an opinion capable of withstanding legal scrutiny.

The IOP’s psycho-legal opinion does not emerge from algorithmic models, templates, or formulaic matrices. It is an act of evidence-based practice informed by grounded research, requiring interpretive reasoning anchored in the claimant’s verified profile, assessed sequelae, and the IOP’s understanding of labour market behaviour and vocational development.

Each opinion is shaped by the unique constellation of factors present in the individual case, substantiated not merely by reference to data, but by the integration of that data through discipline-specific judgment.

The IOP’s task is not to “match” pre- and post-incident earning profiles, but to interrogate how an incident affects a person’s potential to access, retain, and progress within the labour market.

This requires both insight and contextualisation: assessing the impact of medically confirmed impairments, educational disruption, and psychosocial setbacks – without overstepping into diagnostic territory – while drawing on models of work adaptability, skills transference, and market alignment.

The value of the IOP’s opinion lies not in the data alone, but in how that data is interpreted through a work psychology lens. Given the unique characteristics of the South African labour market, I am unconvinced that any AI platform, however advanced, can replicate the judgment required to determine whether a once-viable vocational path has been foreclosed, compromised, or remains realistically accessible.

The IOP’s conclusion is not a procedural output – it is the outcome of deliberate, case-bound reasoning. That reasoning is anchored in expertise, shaped by complexity, and bounded by ethical and legal accountability. It is this interpretive centre, not any procedural framework, that renders the IOP’s role in psycho-legal matters both unique and irreplaceable.

The risk of oversimplification: Why AI cannot capture organisational-psychological complexity

The promotion of AI as a replacement or enhancement for core expert functions of the IOP fosters a dangerously reductive view. It ignores the nuanced, individualised nature of psycho-legal analysis, and the ethical and evidentiary standards it must meet.

The IOP does not simply align earnings data with employment statistics. Rather, each case requires analysis of pre-incident vulnerabilities, contextual realities, and functional limitations as verified by appointed professionals. Human insight remains indispensable.

Human interpretation over algorithmic synthesis

AI can generate resource lists and summarise content, but lacks access to the paid, verified sources required for earning projections. More critically, it lacks the professional discernment to weigh inconsistent evidence, flag psychosocial red flags, or reason within the legal test of probability.

For example, no AI system can:

• Interpret vague or contradictory collateral testimony;
• Evaluate the long-term impact of impairments on job demands;
• Distinguish between stated vs. observed effort;
• Apply independent judgment anchored in professional standards.

AI-generated content often suffers from target fixation, where early inputs overly shape outcomes, thereby amplifying rather than reducing bias. This is directly at odds with the objectivity demanded of expert witnesses.

Evidence-based practice ≠ data processing

In line with the principles of evidence-based practice (American Psychological Association, 2006; Canadian Psychological Association, 2012), a defensible psycho-legal opinion by an industrial and organisational psychologist must reflect the deliberate integration of three core elements: the best available research evidence, the practitioner’s clinical and professional expertise, and the unique needs, values, and contextual characteristics of the individual case.

This is not a data aggregation exercise. It is interpretive, reflective, and rooted in applied psychology.

Applied judgment, not aggregated input: The IOP’s domain of interpretation

The IOP’s contribution lies not in describing facts but in interpreting them through the lens of industrial/organisational psychology. Medical records, employment history, and educational outcomes do not, in isolation, determine post-incident earning capacity. The IOP must assess how these elements intersect to inform realistic vocational trajectories.

This is not linear modelling – it is a forensic act of reasoning. The IOP must weigh what the claimant was occupationally poised to access before the incident, how that trajectory has been altered by functional and contextual disruption, and where possibility gives way to constraint. This is judgment – not algorithm.

This evaluative function is not interchangeable with data processing or generalist commentary. In Ndebele v Minister of Police [2022], the court rejected the evidence of a self-described “earnings expert” whose report mirrored the function of an industrial psychologist but lacked professional registration, methodological rigour, and alignment with professional scope. The court found that her input was neither objective nor helpful, stating that she “failed to provide independent assistance... by way of an objective and an unbiased opinion” (para 70).

The judgment underscores a crucial principle: psycho-legal opinion must emerge from within a recognised professional framework. It must reflect the application of discipline-specific judgment to a verified factual base, not unsupported assumptions, heuristics, or opinions beyond scope. That standard cannot be met by algorithmic tools, unregulated contributors, or template-driven logic.

An IOP does not diagnose, but interprets how impairments affect career resilience, adaptability, and access to employment sectors. This analysis cannot be done by software, actuarial tools, or generative AI.

To suggest otherwise misunderstands both scope and liability. The IOP’s role is to answer: How has this incident changed this person’s ability to earn? That is applied judgment – discipline-bound, accountable, and irreplaceable.

In Part 2 of this article, we turn to the ethical, legal, and evidentiary guardrails that govern psycho-legal practice - and examine why these safeguards place clear, enforceable limits on the use of AI in forming expert opinion.

This article is authored by an IOP who performs such work under formal HPCSA registration, subject to statutory and ethical requirements, and in line with court-defined expectations of objectivity and reliability as defined by court precedent.