AI is making something increasingly difficult for organizations to ignore: having knowledge is not the same as being able to use it.
A lawyer may understand why a particular contractual position matters. A compliance professional may recognize an exception that a written policy does not anticipate. An experienced practitioner may know why a process repeatedly fails despite appearing sound on paper. That expertise can be invaluable, yet difficult to transfer because it depends on experience, context and judgment that have never been fully documented.
AI creates a reason to confront this problem. As technology becomes capable of processing information and performing work at scale, organizations have to examine not only what their systems contain, but also the professional judgment on which their work depends.
The question is no longer simply whether AI can perform a task. It is whether an organization has made the knowledge behind that task accessible enough to use well.
The Knowledge Was Already There
Early in my legal career, I encountered a client that wanted its contracts to be as aggressive as possible because, in the client’s view, nobody on the other side was reading the terms closely anyway.
What stayed with me was what this revealed about the agreements themselves. The obligations were there. The risks were there. Yet the people who had agreed to those terms were not necessarily understanding what they meant in practice. The information was not missing. It was inaccessible in the form in which the organization needed it.
A signed contract can contain information that matters long after negotiations end. Commercial teams need to understand existing commitments in order to fulfill them. Business leaders need visibility into obligations that could affect their decisions. Legal teams need to identify risks before they become disputes.
AI also introduces the opportunity to identify and surface tacit knowledge about your organization that may not be explicitly understood. For example, AI introduces the opportunity to monitor how often counterparties negotiate certain terms and identify how such negotiations impact deal cycle times and therefore revenue.
AI can help make information more accessible, allowing contractual terms and trends to inform decisions rather than remain buried in documents. But most of the underlying knowledge was already there. The technology changes the conditions under which people can find and use it.
This distinction extends beyond contracts. Organizations accumulate decisions, standards and exceptions over years of work, but much of that knowledge remains dependent on the people who remember why a particular decision was made or when a rule no longer applies.
The challenge is not simply to collect more information. It is to make existing knowledge useful beyond the circumstances in which it was first acquired.
What Can Be Automated, and What Still Requires Judgment?
As AI becomes more capable, it is tempting to imagine professional work as a gradual transfer of decisions from people to machines. A more useful question is which parts of that work can be standardized and where human judgment remains essential.
AI can extract terms across thousands of contracts, review proposed changes and identify language that falls outside an established standard. These tasks are well suited to automation when the organization has clearly defined what it wants the system to recognize.
The difficulty begins when a situation does not fit the standard. A clause that appears unusual might be justified by the commercial relationship. An exception might reflect a change in the business that the original rule never anticipated. A decision might depend on context that cannot be reduced to a checklist.
An experienced professional’s value lies partly in recognizing these distinctions and exercising judgment around them. They understand not only what a standard says, but when it applies, when it does not and what might justify departing
This makes the quality of the standard itself crucial. Consider a negotiation in which two lawyers disagree about what constitutes a market-standard clause. A more powerful AI model cannot resolve a disagreement the organization has never settled. People must first decide what they consider acceptable, where the boundaries lie and when an exception is justified.
The same principle applies to trust. When a system flags a potential problem, the person reviewing it needs to understand the reasoning or standard behind the result. An answer is not reliable simply because a machine produced it.
AI can apply established judgment at scale, but organizations remain responsible for defining that judgment, examining its limitations and deciding when a person needs to intervene.
Why Expertise Does Not Travel
If professional judgment is so valuable, why do organizations leave so much of it with individual people? Incentives are part of the answer. Organizations often ask experienced professionals to share what they know without giving them a meaningful reason to document it.
A lawyer who has spent fifteen years developing a sophisticated approach to contract language may possess knowledge that could improve the work of an entire team. Yet the organization may have no clear mechanism for recognizing or rewarding the effort required to turn that experience into something others can use.
The knowledge consequently remains in someone’s head, inside a firm’s templates or within a small group of people who understand how the work really gets done. The organization benefits from their expertise, but access to it depends on finding the right person and having the opportunity to consult them.
This is not merely a documentation problem. It is also a question of how organizations value knowledge and the people who develop it.
The creator economy offers an interesting comparison. In fields such as teaching, coding and design, people have developed ways to turn expertise into work that others can access, use and build upon, sometimes receiving recognition or compensation when that work creates value.
Professional standards could benefit from a similar principle. Experts need ways to articulate what good work looks like, explain the reasoning behind their decisions and make those principles available for others to apply.
AI makes this possibility more practical. Once a judgment is expressed clearly enough to inform a standard, it can potentially be applied repeatedly, reviewed by others and improved over time. The person who developed it no longer needs to be personally involved in every instance.
That does not eliminate the need for expertise. It changes how expertise can contribute to the organization.
What Happens to the Expert’s Role?
When professional knowledge becomes easier to share, the role of the expert begins to change.
Traditionally, an organization might send a difficult contract to its most experienced lawyer because that person had the knowledge needed to assess it. Their time and availability limited how many matters they could personally review.
If the principles behind their decisions can be made explicit and incorporated into a reliable standard, that expertise can inform many more decisions. The professional can spend less time repeating established work and more time examining situations where the right answer is genuinely uncertain.
Legal and compliance professionals may therefore find themselves spending more time defining acceptable risk, deciding when a standard should be revised and determining whether an exception makes sense in the circumstances.
This is why AI should not be treated purely as a productivity story. Processing more contracts or completing reviews faster may be valuable, but those gains do not explain what the organization does with the capacity it has created. The more consequential question is whether people use that capacity to make better decisions or create value in other ways.
Leadership influences whether that process becomes an opportunity for learning or a threat to professional authority. When leaders experiment openly, acknowledge uncertainty and demonstrate a willingness to learn, they make it easier for others to do the same.
Organizations must also learn to ask experienced professionals questions they may have answered intuitively for years. What makes a particular risk acceptable? Why does one exception matter more than another? Under what circumstances should a standard be reconsidered?
Capturing those answers requires more than introducing new technology. It requires making the reasoning behind professional decisions an explicit part of how work is done.
What AI Cannot Invent
AI readiness is often approached as a question of choosing the right model, connecting the right systems and identifying suitable tasks for automation. Those decisions matter, but they cannot compensate for knowledge that an organization has never clearly established.
Before automating a process, it is worth asking where the relevant knowledge resides, which standards people actually agree on, how exceptions are handled and what happens when an experienced person leaves. These questions reveal whether the organization can explain the decisions it makes or relies on individual memory to keep its processes working.
There is a limit to what technology can accomplish here. AI can help surface information, analyze implied standards and reduce repetitive work. It cannot automatically turn incomplete reasoning into sound judgment, nor can it guarantee that a rule developed for one situation will be appropriate in another.
The task is therefore not simply to capture everything experienced professionals know. It is to identify which parts of their knowledge can be articulated, tested and shared, while preserving human judgment where context still matters and uncertainty is high.
That distinction has implications for the future of professional expertise. If knowledge can be made available beyond the person who developed it, expertise becomes less dependent on individual availability. But the value of the expert does not disappear. It shifts toward defining sound standards, monitoring the quality of the AI’s application of those standards, and handling exceptions or edge cases.
The more organizations automate, the more important it becomes to understand what their technology is applying, whose judgment shaped it and when that judgment needs to be reconsidered.
Perhaps the defining question of AI in professional work is not whether machines can replace experts. It is whether organizations can make better use of what their people know without losing the judgment that makes that knowledge valuable.
