AI can draft the memo — but can it teach judgement?

As AI strips away the formative elements of junior legal training, we must deliberately cultivate virtues that were once absorbed through slow, imperfect human training

What, if anything, do we need to change about our legal education and training, ie, what skills will students/the newly qualified need for the future?

The question asks about skills. That is the right place to start, but it is not the only question.

Legal practice has always required two different capacities: tasks and purposes. Tasks are activities directed at outputs: drafting, research, review. Purposes are what those tasks are for: understanding the client’s real situation, judging well under uncertainty, remaining answerable for where your advice leads.

Tasks require skills. Purposes require something else. They require what I call virtues, capacities of character, not technique.

AI automates tasks. It does them well and it will do them better. So, yes, students and juniors will need new skills. AI literacy. The ability to evaluate and challenge AI outputs. These are important and they are teachable.

But here is the harder point. The traditional training model never separated skills from virtues because it never needed to. The junior who spent three days on a research memo was not just learning to research: she was learning to sit with uncertainty, to notice what was missing, to tolerate not knowing the answer yet. The junior whose draft came back covered in annotations was not just learning to draft better. She was learning what mattered – not just how to draft, but what the draft was protecting.

The tasks were the medium. Nobody designed this. It worked by legal osmosis – the assumption that training skills would cultivate virtues as a byproduct.

AI breaks that assumption. When the task is automated, the formation space inside it disappears.

Virtue by design

So what needs to change is not only the skills list. It is the recognition that we now need to cultivate virtues deliberately because the work will no longer do it for us.

That runs through the whole pipeline. At law school, we need to form virtues through hard cases and moral complexity, not ethics as a compliance module. In training, we must protect the slow, inefficient occasions where judgment is transmitted: the observed client meeting; the corrected draft; the conversation about why, not just how. At partner level, we need to recognise that senior lawyers are not just supervisors but moral exemplars, and that this role cannot be delegated to a platform.

The question asks what skills students will need. The honest answer is: better ones than before.

But the deeper answer is this. We must stop assuming that training skills will cultivate virtues. It used to. It will not any longer. And unless we build that cultivation deliberately into every stage of the pipeline, we will produce a profession that performs tasks brilliantly but has forgotten what they were for.