The short answer. AI will not replace lawyers, but it is already taking over parts of the job: first-pass document review, contract summaries, research leads and early drafts. The US Bureau of Labor Statistics still projects 5% job growth for lawyers from 2025 to 2035. What changes is how lawyers spend their hours and what they must check.
If you've tried a generative AI tool on a legal task, you've seen how fast a first draft appears. The real question is what that speed does to the job.
Here is what the evidence says, task by task, along with the rules that keep lawyers accountable for every word AI helps them write.
What does AI already do in legal work?
AI already handles much of the high-volume, first-pass layer of legal work. The American Bar Association's Formal Opinion 512 says generative AI tools "may assist lawyers in tasks such as legal research, contract review, due diligence, document review, regulatory compliance, and drafting letters, contracts, briefs, and other legal documents." In every case, the output is a starting point the lawyer still owns.
| What AI does today | What stays with the lawyer |
|---|---|
| Document review in discovery. Sorts large document sets as responsive or not and flags likely privileged material | Privilege calls, production strategy, and what the findings mean for the case |
| Legal research. Surfaces cases and statutes and drafts research summaries | Confirming every authority exists and says what the summary claims |
| Contract review and due diligence. Summarizes and compares long contracts | Deciding which risks matter to this client and what to negotiate |
| Drafting. Produces first drafts of letters, contracts and briefs | The argument, the facts, and responsibility for what gets filed |
| Client advice and negotiation. Prepares outlines and talking points | The work itself: the ABA says it cannot be left to AI alone |
The ABA's wording on that last row is direct. Lawyers may not leave it to AI tools alone "to offer legal advice to clients, negotiate clients' claims, or perform other functions that require a lawyer's personal judgment or participation."
The weak point is accuracy. Stanford researchers found that leading AI legal research tools each "hallucinate between 17% and 33% of the time," even though they did better than general-purpose chatbots. A hallucination is a confident answer with no basis in fact (more on hallucinations). In legal work, that often means a case that does not exist.
Courts have already punished that mistake. In Mata v. Avianca, a federal court in New York imposed a $5,000 penalty on two lawyers and their firm who filed non-existent judicial opinions created by an AI chatbot, then kept standing by them after the court questioned them.
Judge P. Kevin Castel's sanctions opinion (June 22, 2023) put the line in the right place: "there is nothing inherently improper about using a reliable artificial intelligence tool for assistance. But existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings." Read the opinion.
What does the job outlook say for lawyers?
The federal outlook is steady growth. The BLS Occupational Outlook Handbook says: "Employment of lawyers is projected to grow 5 percent from 2025 to 2035, faster than the average for all occupations." It expects about 28,700 openings a year and reports a median annual wage of $159,670 in May 2025.
BLS also speaks to automation directly:
"Some routine legal work may be automated. However, this increased efficiency is not expected to reduce overall demand for lawyers. Instead, automation is expected to allow lawyers to spend more time on other aspects of their work like meeting with clients and strategizing."
That is more useful than either extreme. Routine work shrinks, and the hours that remain tilt toward clients and strategy.
The profession's own rules point the same way. The ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512 on July 29, 2024. Its opening summary lists the duties in one sentence:
"To ensure clients are protected, lawyers using generative artificial intelligence tools must fully consider their applicable ethical obligations, including their duties to provide competent legal representation, to protect client information, to communicate with clients, to supervise their employees and agents, to advance only meritorious claims and contentions, to ensure candor toward the tribunal, and to charge reasonable fees." ABA Formal Opinion 512
The opinion is also plain about the limits. AI tools may "significantly assist lawyers in serving clients," it says, but "they cannot replace the judgment and experience necessary for lawyers to competently advise clients about their legal matters or to craft the legal documents or arguments required to carry out representations."
It touches the business model too. Lawyers who bill by the hour "must bill for their actual time," so work that takes minutes instead of hours changes how firms price, not just how they work.
Which skills keep lawyers valuable?
The lawyers who gain from AI treat it like a fast junior whose work always gets checked. Four skills matter most, and each lines up with a duty in the ABA opinion: verification, clear briefing, confidentiality judgment, and supervising AI-assisted work done by others.
- Verification as a reflex. Check every citation, quote and fact before it leaves your hands. Know why grounding a tool in your own documents, called retrieval-augmented generation (RAG), cuts errors without removing them. The ABA ties this to competence: relying on AI output without "an appropriate degree of independent verification or review of its output" could breach Model Rule 1.1.
- Briefing the model well. Give it the matter context, the audience, what good looks like and an example. That is prompt engineering in plain terms, and it is what turns a generic summary into a usable one.
- Confidentiality judgment. Know which tools your firm has approved for client information and what each does with your inputs. The ABA analyzes this under Model Rule 1.6, the duty of confidentiality. The AI governance track covers the basics.
- Supervising AI-assisted work. When associates and paralegals use AI, the lawyer who supervises them carries the responsibility. The ABA says supervisory duties include making sure lawyers and nonlawyers are trained "in the ethical and practical use of the GAI tools relevant to their work as well as on risks associated with relevant GAI use." The same shift for support staff is covered in Will AI replace paralegals?
These are habits more than facts, which is why they build through regular practice rather than a single seminar. What is AI fluency? explains the difference between knowing about AI and using it well.
Try three real kju questions for lawyers
These three come from kju's daily practice for legal teams, and kju users get eight like these each day, adapted to what they miss.
Question 1. In Legal & Compliance at a law firm, which task is the smartest first thing to hand to an AI assistant?
- A) First-pass review of standard contracts
- B) Sending AI-drafted legal advice without review
- C) Everything at once, to maximize time savings
- D) Nothing: AI isn't reliable enough for any of it
Answer: A. Start where volume is high and mistakes are cheap, and keep review on everything that leaves your hands. Source
Question 2 (true or false). Once an AI assistant is connected to precedents and internal know-how, its answers no longer need human review.
- True
- False
Answer: False. Grounding a model in your documents reduces made-up answers but doesn't eliminate them. Review stays. Source
Question 3. You're using AI to produce a clause-by-clause contract summary. The first output is generic and misses the point. Best next move?
- A) Give it context: audience, what good looks like, an example
- B) Regenerate until it comes out right
- C) Give up: the model can't do this task
- D) Ask it to make the answer longer
Answer: A. Models aren't mind readers: context, a definition of good, and an example beat blind regeneration. Source
What should lawyers do next?
Pick one task you repeat every week, such as a first-pass contract summary, and practice it with AI plus a short verification checklist. Small, daily repetitions are what turn checking into a reflex. A few minutes a day with kju keeps that practice going.
If you lead a practice group or an in-house legal department, kju trains whole legal teams: each person gets daily practice tailored to their role, and you can see progress across the team. See kju for organizations and kju for legal teams.
For the wider picture, start with What jobs will AI replace?, then compare a neighboring profession in Will AI replace accountants?
Frequently Asked Questions
- Will AI replace lawyers in the next 10 years?
- Unlikely. The US Bureau of Labor Statistics projects employment of lawyers to grow 5% from 2025 to 2035, with about 28,700 openings a year. It expects some routine legal work to be automated, but says that efficiency is not expected to reduce overall demand for lawyers. The job changes more than it disappears.
- Can AI replace lawyers in court?
- No. AI can help prepare a filing, but the lawyer stays responsible for it. In Mata v. Avianca (2023), a federal judge imposed a $5,000 penalty on lawyers who filed fake case citations created by an AI chatbot, noting that attorneys have a gatekeeping role to ensure the accuracy of their filings.
- What does ABA Formal Opinion 512 say about AI?
- Issued on July 29, 2024, it says lawyers using generative AI must consider their duties of competence, confidentiality, client communication, supervision, meritorious claims, candor toward the tribunal and reasonable fees. It also says AI tools cannot replace the judgment and experience lawyers need to advise clients.
- Which legal tasks is AI most likely to automate?
- High-volume, first-pass work: document review in discovery, contract review and summaries, due diligence, research and first drafts. The ABA lists these among the tasks generative AI may assist with. The output still needs a lawyer's review, because AI research tools can cite authority that does not exist.