News & Views

AI vs Professional Judgment: the False Economy Businesses Cannot Afford

Artificial Intelligence (AI) is rapidly becoming a cornerstone of modern business strategy, with many organisations deploying it as a powerful cost-control lever by automating routine work, compressing timelines, and driving operational efficiency at scale. For professional services firms, the commercial rationale is compelling: AI offers speed, scalability, and accessibility, presenting (to a certain extent) a cost-effective means of addressing day-to-day operational and advisory demands.

The rapid adoption of these generative tools has supported businesses in producing draft documents in seconds, synthesising large volumes of information, and supporting internal decision-making with minimal marginal cost. Increasingly, however, business leaders are extending this use case beyond efficiency gains, and are now testing the boundaries of AI’s role in areas traditionally reserved for qualified professional judgement.

A clear trend is now emerging in which organisations are turning to AI-generated outputs to inform decisions in areas such as legal interpretation, financial planning, insurance coverage, and regulatory compliance, often without engaging qualified advisers or subjecting those outputs to appropriate professional review.

This shift gives rise to a more consequential question: not simply whether AI can replicate the judgement, contextual sensitivity, and accountability inherent in professional advice (which is a growing concern among UK regulatory bodies) but whether its increasing use in these contexts is introducing a new category of risk. What then, are these risks?

One of the most widely cited concerns is the risk of hallucinations, i.e., outputs that are fluent and persuasive, but factually incorrect or entirely fabricated. In a legal context, this can manifest as invented authorities, mischaracterised statutory provisions, or confidently stated conclusions that have no grounding in the applicable framework.

There’s also a more structural risk in the use of AI-generated contractual documentation where outputs omit key protections, misallocate risk, or fail to reflect the commercial intent of the parties. This is particularly evident with more nuanced judgment-driven provisions which may be absent or inadequately framed, resulting in agreements that appear robust but ultimately fail when tested.

More concerningly, reliance on AI-derived guidance in decision-making is creating both immediate and lasting exposure. While misinterpretation of regulatory obligations, flawed financial assumptions, or incomplete risk assessments may initially go unnoticed, they’ll often crystallise into disputes, regulatory sanctions, and commercial loss at a later stage. The key danger here lies not in the existence of these errors or omissions, but, in the absence of professional guidance, that they’ll continue to remain undetected and will then be relied upon in the process of informed decision making.

This isn’t to suggest that professional advice is immune from errors. No system, human or artificial, is infallible, but the distinction lies in accountability and protection. Where a professional adviser is engaged, any advice provided is supported by a duty of care and, in most cases, backed by professional indemnity insurance. That framework exists to protect the client if something goes wrong and provide suitable means of recourse to protect their business and reputation. AI in contrast offers no equivalent safeguards, and the risks, as well as all costs that are incurred in bearing them, rest entirely with the business that opts to rely entirely on AI-generated outputs.

Against this backdrop, the commercial reality becomes difficult to ignore: using AI as a substitute for professional input doesn’t remove the need for that input, it only defers it. Those same issues that go undetected at the point of decision-making will typically resurface once they’re tested, by which stage they’ve often escalated into disputes, regulatory scrutiny, or measurable financial loss, and are no longer matters of optimisation, but remediation. Organisations are then left to unwind decisions, renegotiate or reconstruct arrangements, and engage professional advisers under time pressures and with reduced optionality. Crucially, the cost of taking these remedial steps, both financially and operationally, will frequently exceed what would have been incurred had appropriate advice been sought from the outset.

For businesses, the position is straightforward: professional advice can be taken now, as part of structuring an outcome, or it will be required later, as part of unpicking one. The only real variable is cost.

For further information please contact Michael McKenna or call 0151 906 1000.