# What AI skills do junior lawyers actually need?

Source: https://perfraction.com/blog/ai-skills-for-junior-lawyers/
Author: Joshua Woo, Perfraction
Published: 2026-09-16
Topics: Legal profession, Legal tech, AI skills

## Short answer

Not prompt engineering. Junior lawyers need to understand how AI actually works, use that to understand how it fails, and build checks that catch their own blind spots, including the embarrassment test of whether they could defend the output to a partner out loud.

## Key takeaways

- AI is predictive, not sentient. Treat its output as a bright intern's first draft.
- Hallucination includes misreading what is there and missing what isn't, not only inventing clauses.
- Grounding checks and adversarial cross-reviews catch errors a single read won't.
- The real risk isn't AI replacing judgement. It is lawyers switching their own off.
- Law school and the bar don't test these skills yet.

## Why isn't prompt engineering the answer?

Someone asked me what skills a junior lawyer or law student needs to thrive today. I had just presented, with my team, a solution on apprenticeship and attrition in junior legal practice at the judging round of Hackathon for a Better World 2026, co-organised by the Singapore Academy of Law, DBS Bank and the Singapore Courts.

After months of building AI legal workflows, my answer lands on three things. None of them is prompt engineering.

## 1. Understand how AI actually works

It is predictive, not sentient. It has no common sense, however fluent it sounds. Treat its output like a bright intern's first draft, not a senior partner's final version. Question everything by default.

## 2. Let the "why" tell you the "how"

Once you understand how it works, you understand how it fails. Hallucination isn't only inventing a clause that doesn't exist. It is misreading what's there, and missing what's not.

## 3. Build the checks that catch your own blind spots

- Grounding checks against the source.
- Adversarial cross-reviews.
- The embarrassment test: could you defend this output to a partner, out loud, right now?

The profession's real risk was never AI replacing judgement. It is lawyers quietly switching their own off.

## Where do these skills get taught?

Not yet in law school or the bar exams. They will matter more than most of what does get graded.

## Frequently asked questions

### Do lawyers need to learn prompt engineering?

It helps a little, but it isn't the core skill. Knowing how AI fails and checking its output matters far more.

### What is the embarrassment test for AI output?

Before relying on AI output, ask whether you could defend it to a partner, out loud, right now. If not, check it further.

### What is a grounding check?

Verifying each statement in AI output against the source document or authority it claims to rely on.

---

This article is general information, not legal advice. It reflects the position as at the date of publication.

About Perfraction: a Singapore-based fractional legal function for technology companies, led by Joshua Woo. Perfraction is a legal consultancy and not a law firm. We do not provide legal advice or legal representation. Our services are limited to strategic advisory, legal operations, and regulatory consulting. For legal advice, please consult a qualified lawyer or licensed law firm.
Contact: josh@perfraction.com · Book a call: https://calendly.com/perfraction/30min · Questionnaire: https://perfraction.com/questionnaire