Previously, I explored why attorneys may be more willing to seek substantive critique from AI than from another person. AI creates a different dynamic, and it offers a sounding board without the immediate interpersonal considerations that can sometimes make feedback difficult to seek, deliver, or fully absorb. Yet, the developmental core of feedback still requires human judgment, context, trust, and an understanding of both the work and the individual trying to improve it.

The opportunity, then, is not simply to use AI to provide more feedback; rather, it’s to use what we are learning from these interactions to improve the complete feedback loop. This includes the conversations that should happen before AI is used, the analysis of what it produces, and the human coaching that follows.

Start with the work and the purpose

One reason AI feedback may feel easier to engage with is that it tends to keep the immediate focus on the work product. A user can prompt AI about whether an argument is clear, whether a draft accounts for the intended audience, or whether an analysis is missing a counterpoint. The exchange is specific and iterative, and the individual retains discretion to question, refine, or reject the response.

There is something useful in that structure for legal teams, whether in a law firm or an in-house legal department. Feedback is often most productive when the purpose of the assignment is clear, and the conversation remains grounded in what the work needs to accomplish. For example, what is the client or business trying to solve? Who is the audience? What risks, strategic considerations, or practical realities should shape the answer? Where would another perspective strengthen the analysis?

Feedback is often most productive when the purpose of the assignment is clear, and the conversation remains grounded in what the work needs to accomplish.

AI can help bring some of those questions to the surface, but the supervising attorney, partner, or subject-matter expert must supply the context that gives the answers meaning. That includes knowing whether the AI’s suggestion is legally sound, strategically relevant, consistent with the organization’s values and quality standards, and appropriate for the specific client or business situation.

Build feedback into the assignment

Ideally, the feedback loop begins when the assignment is given rather than after a completed draft has already missed the mark.

A junior attorney may review an assignment and identify where AI could be useful. Before proceeding, the junior and supervising attorney should discuss the proposed approach, which approved tool may be most appropriate, what instructions or prompts should be provided, and what the tool is being asked to accomplish. Reviewing the AI output together then creates another opportunity for substantive development. What did it identify? What did it overlook? Which recommendations should be incorporated, and which should not? Most importantly, why?

That final question is where much of the learning happens. A junior lawyer can learn the technological how while also developing the substantive why behind legal reasoning. The supervising attorney also benefits by seeing the tool’s capabilities and limitations directly, rather than simply telling someone to use it and waiting for a finished output.

Indeed, this is one way the feedback loop can promote simultaneous development across experience levels. And at Seyfarth, this reflects the philosophy behind SEYmultaneous Advancement, our firmwide approach to have our people learn with and from one another as AI becomes more embedded in legal work.

This current R&D mode of experimentation is happening across many organizations, but typically it’s in siloes or on an individual basis. However, the most valuable moments occur when one person shares an approach and someone else questions the process or suggests alternate methods. In this way, ideas improve out in the open, and others can then adapt them to their own role, practice, department, or workflow.

Connect developmental feedback to formal evaluation

AI also gives legal organizations an opportunity to examine whether their day-to-day feedback systems connect meaningfully to more formal performance conversations.

A performance review should not be the first time someone learns that an aspect of their work needs to improve. To mitigate that from happening, AI can help support more regular developmental touchpoints throughout the year. Also, it can help an attorney identify recurring questions, prepare for a coaching conversation, or compare how a draft evolved after several rounds of human and AI input.

This does not mean using AI to determine a performance rating or allowing generated observations should become a proxy for a manager’s own judgment. Formal evaluation involves the use of context that an AI system does not possess, including consideration of the complexity of assignments, as well as the junior lawyer’s progress over time, contribution to teams, professional judgment, and client service. Instead, AI can help create more occasions for reflection before the formal review takes place, reducing the likelihood that feedback arrives late or feels disconnected from the work that the junior produced.

Turning the AI paradox into an opportunity

What piques my interest in particular is not simply that AI can provide another source of feedback. It is what our willingness to engage with that feedback may allow us to do differently in developing legal talent.

For years, firms and legal departments have been looking for ways to make feedback more timely, more continuous, and more connected to the work itself. AI can help create more opportunities for that to happen in the natural flow of an assignment instead of at the end of a project or in a formal performance review. AI also can give an attorney a way to test an idea and gain an initial perspective to consider before bringing the work into a conversation with someone who can provide the context and judgment the technology cannot.

AI also gives legal organizations an opportunity to examine whether their day-to-day feedback systems connect meaningfully to more formal performance conversations.

The important point is that AI interaction should not become the endpoint, but rather a new entry point into mentorship. A junior attorney who has already explored an issue with AI may be better prepared to ask more substantive questions, and the supervising attorney can use that same exchange to teach legal reasoning, judgment, client context, and the limitations of the tool, while learning more about how AI is being used in the actual workflow.

I do not think the future of feedback is necessarily strictly human or AI. The real opportunity is to bring the two together thoughtfully, using AI to create more openings for learning and using human mentorship to turn those openings into meaningful development.

If legal organization are able to do that well, AI will not distance attorneys from the apprenticeship and relationships that are at the core of the profession; rather, it may actually help them make those relationships more deliberate and continuous — and ultimately more valuable.

You can find out more about the impact of AI and other advanced technologies on the legal profession here‍