If AI Can Do the Technical Work, What Makes a Great Lawyer?
As artificial intelligence becomes increasingly capable of performing complex technical tasks, our understanding of professional expertise may need to evolve with it.
Nvidia CEO Jensen Huang recently offered an interesting perspective on this shift. Asked to identify the smartest person he had ever met, Huang instead questioned the conventional definition of intelligence itself. Technical ability and problem-solving, he argued, are becoming increasingly commoditised as AI systems learn to perform tasks that were once considered evidence of exceptional intellectual ability. Software programming is one example: long regarded as a highly specialised skill, it is already being transformed by generative AI.
Huang's alternative definition of intelligence is broader. He describes the most valuable people as combining technical expertise with empathy, experience and an ability to infer what is not immediately apparent — people who can, in his words, "see around corners".
For the legal profession, this raises a compelling question. If AI can perform more of the technical work traditionally associated with expertise, which human capabilities become more valuable?
AI and the legal profession: from information to judgment
The impact of AI on the legal profession is already visible. Generative AI can assist with legal research, summarise extensive documents, compare contractual provisions, analyse evidence and produce first drafts of legal material. These capabilities are likely to improve further.
This does not make legal expertise redundant. Rather, it may change where that expertise matters most.
One observation Huang makes elsewhere in the interview is particularly relevant. As computers become capable of tackling increasingly complex problems, the challenge is not necessarily limited to finding the answer; there is also the question of how the problem should be formulated in the first place.
Law provides a good illustration. A dispute may involve thousands of documents, several plausible legal arguments and competing accounts of the same events. AI can help lawyers navigate that information, but determining which issues deserve attention requires something more than efficient information processing.
The apparent contractual issue may conceal a more significant question of authority or credibility. A legally sound argument may be strategically unhelpful. A factual inconsistency that appears insignificant in isolation may become important when considered alongside the wider evidence.
As AI makes analysis faster and more accessible, professional judgment may therefore become more, rather than less, important.
Why first-principles thinking matters in legal AI
Huang also places considerable emphasis on first-principles reasoning: examining the assumptions underlying a conclusion and being prepared to revisit them as circumstances change.
This has obvious implications for legal AI. Increasingly sophisticated AI systems can produce fluent and convincing answers, but fluency should not be confused with reliability. Lawyers using these tools still need to understand the assumptions behind an analysis, identify missing information and consider whether an apparently persuasive conclusion survives scrutiny in its broader legal and factual context.
The skill involved is not simply knowing how to obtain an answer from an AI system. It is knowing how to evaluate that answer.
This distinction is likely to become increasingly important as AI tools are integrated into everyday legal practice. Greater access to information does not eliminate the need for expertise; it places greater emphasis on the ability to interpret, question and apply it.
What does it mean to "see around corners" in arbitration?
Huang's idea of being able to "see around corners" is particularly resonant in international arbitration and dispute resolution.
Experienced practitioners routinely make judgments that depend on a combination of legal knowledge, experience and understanding of human behaviour. Counsel may recognise that a minor evidentiary issue is likely to become significant later in proceedings, or that an argument which is technically strong may nevertheless be unpersuasive to a particular tribunal. An arbitrator may identify a commercial or interpersonal dynamic that helps explain a dispute more clearly than the parties' formal positions suggest.
Huang describes this form of intelligence as a combination of data, analysis, first principles, life experience, wisdom and the ability to sense other people. Importantly, technical competence and empathy are not opposites within this definition; they work together.
That distinction matters in dispute resolution, where the material being analysed ultimately concerns people, organisations and decisions made in particular circumstances. AI can identify patterns across enormous quantities of information, but understanding the significance of those patterns may still depend heavily on context.
Human expertise in an AI-enabled legal profession
Debates about AI and lawyers are often framed around replacement: which tasks will be automated and which jobs will disappear?
Huang offers a somewhat different prediction. He argues that it is more likely that virtually every job will change than that large portions of employment will simply disappear.
For law, this may be the more useful question to consider.
Legal research, document review, drafting and knowledge management are already changing, and client expectations are likely to evolve alongside them. As some forms of technical work become easier to perform with AI, the capabilities that distinguish excellent practitioners may also change.
The lawyers best placed to work alongside increasingly capable AI systems may be those who can formulate the right questions, interrogate assumptions, recognise emerging risks and understand the wider human context in which legal problems arise.
AI may make technical intelligence more widely available. The more interesting question for the legal profession is what we choose to do with it.
