The growing reliance on artificial intelligence (AI) in legal work is increasingly dividing opinion between skeptics and believers. Skeptics warn of inaccuracy, ethical compromise and declining service quality. Believers, on the other hand, maintain that AI is not only inevitable but indispensable to modern legal practice. The truth, however, lies somewhere in between. AI is neither a panacea nor a threat to be resisted. It presents a shift in how legal services are being delivered to businesses.

What distinguishes the current wave of AI from earlier legal technologies is the rise of generative AI and large language models. These systems are capable of understanding and producing human language with remarkable fluency. In practical terms, this means that tasks which once consumed significant legal time such as data analysis, large-scale document review and legal research can now be performed faster and at scale.

This efficiency dividend is already visible in dispute resolution. AI tools are now routinely used to sift through vast volumes of documents, extract relevant facts and organise evidence in a manner that allows lawyers to focus on strategy rather than process. More advanced applications go further, using predictive analytics to assess likely outcomes based on historical data. At the far end of this spectrum lies automated dispute resolution, where entire claims can be processed through online platforms that guide parties from filing to resolution with minimal human intervention.

For businesses, this evolution presents a compelling proposition. Disputes can be resolved more quickly, at lower cost and without the procedural complexity that has at times defined traditional modes such as litigation and arbitration. Indeed, global platforms already resolve low-value disputes through automated systems, with human oversight reserved for more complex matters.

Yet it is precisely at this point that the skeptics’ concerns become more persuasive. Dispute resolution is not simply a mechanical exercise in applying legal rules to data. It involves context, judgment and, often, an appreciation of human behaviour and motive. AI, for all its capabilities, operates on patterns and probabilities. It does not understand nuance in the way a human decision-maker does. This limitation becomes particularly significant in complex commercial disputes, where outcomes often turn on qualitative factors that cannot easily be reduced to data. These concerns also bring to the fore issues surrounding the unauthorized practice of law, particularly where AI tools are used to generate court documents which, should only be prepared by qualified Advocates.

Recent decisions from Kenyan courts illustrate this particular tension. In one instance, court documents were struck out on the basis that they were generated using AI and failed to meet substantive and procedural requirements for court documents. In another case, the use of AI was viewed as conferring an undue advantage on one party. Similarly, in the US, courts have increasingly sanctioned and fined lawyers for using AI to prepare court documents containing fabricated case citations and quotations. These decisions, though still emerging, signal a judicial unease with how AI is being deployed indiscriminately without regard to substantive safeguards. Together, they reflect a system grappling with a new technology increasingly being deployed without sufficient restraint.

AI systems are known to produce inaccurate outputs, sometimes referred to as hallucinations, where responses are plausible but incorrect. There are also concerns around bias in training data, confidentiality of client information and the question of liability when AI-generated content proves erroneous. In a profession built on precision, confidentiality and accountability, these concerns go to the heart of legal practice, with direct implications on the businesses they advise and represent.

However, these risks are not without mitigation. Techniques such as requiring AI systems to cite sources, grounding outputs in verified data and maintaining strict human oversight can significantly reduce error. Nevertheless, the responsibility for the final product remains with the lawyer.

Perhaps the skeptics are correct in stating that AI use needs to be minimized and disputes should be resolved by human beings, however, the believers are not wrong in stating that the increased use of AI is inevitable. Administrative and preparatory functions such as document review, legal research and case organisation are increasingly accepted and pose minimal threat to the integrity of proceedings. However, tasks such as drafting legal documents or analysing evidence should not be substituted by AI use.

The future of dispute resolution therefore lies not in replacing lawyers or judges with AI, but in redefining its role. AI will handle the repetitive administrative tasks allowing practitioners to focus on strategy, advocacy and judgment. In this sense, the most effective model is not substitution but collaboration in a system where human expertise and machine efficiency complement each other. This shift is already influencing client expectations. Businesses are no longer asking whether AI can be used, but how it can be used to enhance efficiency and reduce cost without compromising quality or confidentiality. Law firms that fail to respond to this expectation risk falling behind, not because AI will replace them, but because others will use it more effectively.

At the same time, regulation is beginning to take shape globally, with jurisdictions adopting risk-based approaches to ensure that AI systems are used responsibly. Kenya’s AI legislation should follow a similar path. The challenge for policymakers will be to strike a balance between encouraging innovation and safeguarding the fundamental principles of fairness, transparency and accountability.

Ultimately, the debate on AI in legal practice is not a binary one. The skeptics are right to caution against its unrestricted adoption. The believers are equally right to recognise that AI is here to stay. The more important question is how to integrate it in a manner that enhances, rather than undermines, the administration of justice.

This article was published in the Business Daily  and can be accessed here.