The Rectification of Names in the Age of AI Contracts: Confucius Meets the Automated Legal Brief

When the names we give to things no longer match their true nature, what happens to the systems we build? How does the rise of artificial intelligence in legal drafting test our deepest professional commitments? And where does genuine self-efficacy lie for the modern lawyer in an age of automated brief production?

More than two thousand years ago, Confucius observed that names and reality must correspond. The principle, known as zhengming or the rectification of names, runs through the Analects like a foundational thread. Confucius taught that when names are not correct, speech does not accord with things. When speech does not accord with things, affairs do not succeed. When affairs do not succeed, ritual and music do not flourish. When ritual and music do not flourish, punishments do not fit the crimes. When punishments do not fit the crimes, the people do not know where to place their hands and feet. He was describing not merely a linguistic problem but a systemic one. Language and reality must align. If they separate, chaos follows.

This ancient wisdom speaks directly to a crisis unfolding in legal practice today. Generative artificial intelligence has entered law firms, chambers and compliance departments with remarkable speed. Lawyers now use these systems to draft motions, analyse contracts, summarise case law and produce client advice. The efficiency gains are real and tempting. A junior solicitor working late in a London deal room can instruct an AI system to generate a first draft brief in minutes rather than hours. A barrister preparing for trial can feed documents into an algorithm and receive a distilled analysis. The technology promises to liberate legal professionals from mechanical work and allow them to focus on strategy and judgement.

Yet beneath this promise lies a troubling reality. These systems sometimes generate citations to cases that do not exist. They fabricate quotations from statutes that were never written. They present fabricated legal authorities with the appearance of certainty. The technical term for this phenomenon is hallucination, but the language itself obscures what is happening. These are not innocent errors or minor glitches. They are fundamental misalignments between the words the machine produces and the things those words claim to refer to. The name does not match the reality. A citation that appears to exist does not. A legal principle that seems authoritative has no foundation.

For the lawyer who receives such output and trusts it, the consequences can be severe. A solicitor who presents a fabricated case reference in a compliance memo may mislead her client. A barrister who cites a non-existent statute in court acts on false information. A compliance team relying on algorithmically generated risk analysis may miss actual threats because the system has generated plausible sounding but unfounded legal positions. The chaos that Confucius described emerges not as social disorder but as professional failure and loss of trust.

Why does this matter so deeply? Because the law depends on the correspondence between language and reality more than almost any other human practice. A contract works only if the words used genuinely bind the parties to the obligations those words describe. A statute governs conduct only if the words Parliament enacted actually carry the meaning they appear to carry. A judgement persuades and binds only if the legal reasoning it presents rests on authorities that truly exist. The entire edifice of legal order stands on the rectification of names.

When lawyers began to work with artificial intelligence, many brought assumptions inherited from earlier technologies. Software that manages documents, calculates payments, or tracks deadlines operates in a relatively constrained domain. These systems perform well because they work with well-defined structures and clear outputs. Legal language, by contrast, is subtle, contextual and human. It means what it means because humans have given it that meaning through centuries of practice and interpretation. An algorithm cannot simply generate legal language as though it were producing a weather forecast. Yet many lawyers and law firms have proceeded as though it could.

The problem is compounded by a psychological phenomenon. When a system presents information in a confident, well-formatted way, humans tend to believe it. A brief generated by an AI system arrives with proper headings, citations formatted in house style, and logical flow. It looks like professional work. This appearance of authority makes it harder for a reader to question whether the foundations are sound. Bandura and others studying human self-efficacy have shown that when people feel they understand a system and trust its outputs, they are less likely to exercise critical judgement. The lawyer tells himself that the AI has done the research, so he need not. Self-efficacy becomes outsourced to the machine.

But this represents a profound misunderstanding of where true professional self-efficacy actually lies. Genuine self-efficacy for a lawyer is not the ability to produce documents quickly. It is the capacity to think clearly, to verify facts, to make independent judgements about what the law actually requires, and to take personal accountability for the advice given. This kind of self-efficacy cannot be automated. It cannot be delegated to a machine. It lives in the judgement of the individual professional.

The rectification of names offers a path forward. It demands that lawyers return to first principles. Before using any AI-generated text in a document that will be relied upon, the lawyer must verify every citation, every statement of law, every factual claim. This is not a burden to resent but a fundamental professional duty. The system should be treated as a draft for human review, not as finished work. More importantly, lawyers should cultivate scepticism about machine-generated legal content. Just as a client would not accept advice from a junior lawyer without questioning and checking the reasoning, so lawyers should not accept the outputs of AI systems without rigorous verification.

This practice is demanding. It requires time, attention and intellectual discipline. But it recovers what matters most. It restores the correspondence between the names we use in law and the things those names refer to. It rebuilds the chain of professional accountability that runs from the individual lawyer through to the client and the court. It places human judgement back at the centre of legal practice, where it belongs.

Organisations serious about this challenge are already responding. Some law firms have created protocols requiring human verification of all AI-generated citations before client delivery. Some chambers have established guidelines distinguishing where AI assistance is appropriate, with safeguards for areas where accuracy is critical. These are not attempts to reject the technology but to integrate it responsibly, keeping names and realities in alignment.

The stakes are not small. Legal practice rests on trust. Clients trust lawyers to give them reliable advice. Courts trust advocates to present accurate legal authorities. Society trusts the legal system to operate fairly and predictably. When words no longer correspond to things, when citations dissolve under scrutiny, when algorithms generate plausible falsehoods, that trust erodes. The rectification of names is thus not an ancient philosophical curiosity but an urgent contemporary concern.

So what can we conclude? The rise of generative artificial intelligence in legal practice presents a genuine test of professional integrity and self-efficacy. The solution is not to reject the technology but to integrate it carefully, maintaining rigorous human oversight and refusing to outsource the fundamental responsibility of the lawyer: to ensure that the words he or she presents to a client or a court correspond faithfully to legal and factual reality. In doing so, the modern lawyer honours the ancient wisdom that when names are correct, affairs succeed, and when they are not, chaos inevitably follows.