Will AI Replace Lawyers? What the Evidence Shows About Legal Work

Last Updated: September 2026

No, lawyers are not being replaced by AI. The evidence points to a change in what lawyers spend their time on, not to the end of the job. AI is taking over a share of routine legal tasks, while judgment, advocacy and accountability stay with people who hold a license.

So will lawyers be replaced by AI? Not on current evidence. The more useful question is which parts of legal work are changing, how fast, and what that means for the people doing it.

Key Takeaways

• The US Bureau of Labor Statistics projects lawyer jobs to grow 5% from 2025 to 2035, with about 28,700 openings a year.
• AI exposure is measured in tasks, not whole jobs. Document review and first-pass research are the most exposed parts of legal work.
• Claims that AI scored in the top 10% on the bar exam were inflated. Against licensed attorneys, GPT-4 scored around the 48th percentile.
• Legal AI tools still make mistakes. A Stanford study found leading research tools gave incorrect information in more than 17% of test queries.
• Courts hold the lawyer responsible, not the software. That keeps a licensed person accountable for every filing.
• The role is shifting toward review, supervision and client counsel, which raises real questions about how junior lawyers learn.

Will AI Replace Lawyers?

The short answer is no, and the job numbers support it. The US Bureau of Labor Statistics projects lawyer employment to grow 5% from 2025 to 2035, faster than the average for all occupations.

The same forecast expects about 28,700 openings for lawyers each year, in a world where generative AI is already widely available. If the profession were about to be replaced, the outlook would look very different.

The bigger research estimates point the same way once you read them closely. A Goldman Sachs analysis found that roughly two-thirds of US occupations are exposed to some degree of automation by AI.

For the occupations that are exposed, it estimated that roughly a quarter to as much as half of their workload could be replaced. That describes tasks inside a job changing hands, not the job itself disappearing.

Which Parts of Legal Work Are Most Exposed?

The work most exposed to AI has two features. It involves a lot of text, and it has an output someone can check.

In legal practice, that points to a familiar list:

• First-pass document review in discovery
• Summarizing long contracts, depositions and case files
• Early-stage legal research to find relevant cases
• Drafting standard documents from templates

The table below shows how exposure tends to differ across the main parts of the job.

This matches the wider picture. Pew Research Center found that in 2022, 19% of American workers were in the jobs most exposed to AI. In those jobs, the most important activities may be replaced or assisted by AI.

Much of that exposed work has traditionally gone to junior lawyers and paralegals. That is why the effect on training and staffing matters as much as the effect on headline job counts.

What Parts of a Lawyer's Job Stay Human?

The parts that stay human are the ones where someone has to decide, persuade or take responsibility.

Advising a client means weighing their goals, their tolerance for risk and facts that never appear in a document. Negotiation depends on reading the other side and knowing when to push. Advocacy in court means responding in real time to a judge and to the evidence as it unfolds.

These are not only hard for software. They are also tied to a license. Giving legal advice and appearing in court are reserved for admitted lawyers, and a lawyer answers personally for the advice they give.

That accountability is the real limit on replacement. A tool can draft an argument. It cannot be sanctioned, held liable or disbarred if the argument is wrong.

Can AI Pass the Bar Exam?

This claim is often used as evidence that AI can do a lawyer's job. OpenAI's 2023 technical report for GPT-4 said the model passed a simulated bar exam with a score around the top 10% of test takers.

A 2024 study in the journal Artificial Intelligence and Law found that figure was inflated. The top 10% estimate leaned on results from February exams, where many test takers are repeating after failing in July. Against first-time test takers, the study estimated GPT-4 at around the 62nd percentile.

The gap was wider against people who actually passed. Compared with licensed or license-pending attorneys, GPT-4 came in around the 48th percentile overall. On the essay sections, it was around the 15th percentile.

Essays are the part of the exam closest to real legal writing, which makes that low score the more telling result.

Even a strong score would not settle the question. The bar exam tests whether someone knows enough law to start practicing, not how they handle a client, a judge or a negotiation.

Why Does AI Still Need a Lawyer Checking Its Work?

Because it gets things wrong in ways that are easy to miss.

The best-known example is Mata v. Avianca, a 2023 case in which lawyers filed a brief citing court decisions that ChatGPT had invented. The court imposed a $5,000 penalty and made clear that the lawyers, not the tool, were responsible.

Specialist legal tools do better than general chatbots, but they are not reliable enough to work alone. A Stanford study of legal research tools found Lexis+ AI and Ask Practical Law AI gave incorrect information more than 17% of the time.

Westlaw's AI-Assisted Research did so more than 34% of the time. The same team's earlier study found general-purpose chatbots were wrong on 58% to 82% of legal queries.

Error rates like these mean every AI output that reaches a client or a court needs a qualified person to check it. That checking is legal work in its own right. Our article on what happened when a lawyer relied on AI looks at the consequences in more detail.

How Is the Lawyer's Role Changing?

The clearest change is where the hours go. The Thomson Reuters 2024 Future of Professionals report surveyed more than 2,200 professionals across legal, tax and risk. They expected AI to free up 12 hours a week within five years, and four hours a week in the first year alone.

Time saved on drafting and review does not vanish. It tends to move toward three kinds of work:

• Review and supervision. Checking AI output, spotting errors and deciding what is safe to rely on.
• Client counsel. More time spent on the advice and judgment clients actually hire a lawyer for.
• Oversight and governance. Setting rules for how tools are used, what data they can see and who signs off.

That shift also puts pressure on the billable hour. If a task that took ten hours now takes two, firms and clients have to decide how that work is priced. Different firms are likely to answer that question differently.

What Does This Mean for Clients?

For people and businesses who hire lawyers, the change shows up in three places.

The first is speed. Tasks like reviewing a set of contracts or summarizing a case file can be done faster. That can shorten the wait for routine work.

The second is price. When routine work takes fewer hours, clients have more reason to ask how it is billed. Fixed fees for defined tasks may become easier to negotiate.

The third is risk. A lawyer using AI well should be able to explain how they check its output. It is fair for a client to ask which tools are used, what happens to confidential documents, and who reviews the final work.

What does not change is who is responsible. The advice still comes from a licensed lawyer, and that lawyer still answers for it.

What Does This Mean for People Entering the Law?

For law students and junior lawyers, the risk is less about losing a job and more about how skills get built.

Junior lawyers have traditionally learned by doing high-volume review and research. If AI handles more of that first pass, the profession has to find other ways to build judgment. Supervised review of AI output may become part of the answer, because checking work closely is itself a way to learn.

The skills gaining value are the ones AI cannot supply. These include clear judgment, careful verification, sound ethics and the ability to explain law to a client. Working confidently with AI tools matters too, but mainly so that a lawyer knows when not to trust them.

For a closer look at what these tools can and cannot do day to day, see our guide to AI for lawyers.

Frequently Asked Questions

Can AI replace lawyers?

Not on current evidence. AI can handle first-pass review and research, but it cannot hold a license, give accountable advice or appear in court. Lawyer jobs are still projected to grow 5% from 2025 to 2035.

Will AI replace lawyers in the next ten years?

The official outlook says no. The Bureau of Labor Statistics expects about 28,700 openings for lawyers each year over the decade. The work is changing faster than the number of jobs.

Which lawyers are most affected by AI?

Lawyers whose work is mostly high-volume document review, first-pass research or standard drafting are most affected. Work built on advice, negotiation and advocacy is far less exposed.

Will AI replace paralegals and legal assistants?

Parts of their work are exposed, because much of it is routine and text-heavy. The same pattern applies, though. Tasks shift toward checking, organizing and supporting the lawyer rather than disappearing entirely.

Can AI pass the bar exam?

GPT-4 was reported to score around the top 10% of test takers, but a peer-reviewed study found that figure inflated. Against licensed or license-pending attorneys, it scored around the 48th percentile, and around the 15th percentile on essays.

Is it safe for lawyers to use ChatGPT?

Only with care. In earlier Stanford research, general chatbots were wrong on 58% to 82% of legal queries. Lawyers need to verify every citation and avoid entering confidential client information into tools that do not protect it.

What happens if AI gets a legal citation wrong?

The lawyer is responsible. In Mata v. Avianca, a court penalized lawyers $5,000 for filing a brief with cases invented by ChatGPT. Courts treat AI errors as the lawyer's errors.

Will AI make legal services cheaper?

Possibly for routine work. If tasks take less time, the price of that work may fall or shift away from hourly billing. Complex advice and litigation are likely to stay expensive because they depend on judgment.

What skills will lawyers need as AI grows?

Judgment, verification, ethics and clear communication with clients. Knowing how AI tools work matters mainly so a lawyer can tell when an output is wrong.

Executive Summary

The evidence does not support the idea that AI will replace lawyers. US lawyer employment is projected to grow 5% from 2025 to 2035. The research on AI exposure describes tasks changing inside jobs, not jobs disappearing.

The most exposed work is high-volume review, research and standard drafting. The least exposed is advice, negotiation and advocacy.

Legal AI tools still make frequent mistakes, and courts hold lawyers responsible for them, which keeps a licensed person at the center of the work. The real change is in how lawyers spend their time and how new lawyers learn.

What Should You Do Next?

If you are a lawyer, look at how your own week splits between routine tasks and judgment work. That split says more about your exposure than any headline forecast.

If you run a firm, the useful question is where AI fits the work you already do. Our page on AI for law firms covers how that is approached.

People Also Read

• Building an AI Governance Strategy That Scales
• How Do You Implement AI for Small Business Practically?

About the Author

Jerry Jariwalla is the founder of AiBuildrs and creator of the Growth Signal Intelligence framework. With over 22 years in digital marketing and multiple successful business exits, Jerry has spent the past decade leading AI implementation programs for mid-market businesses across professional services, recruitment, membership organizations, and traditional industries.

Expertise: AI Strategy, AI Implementation, Workflow Automation, Custom AI Development, Voice AI, Offshore Engineering, B2B Sales Intelligence, Mid-Market AI Adoption

Connect: LinkedIn

Disclaimer: This content is for informational purposes only and does not constitute professional business or technology advice. ROI outcomes vary based on industry, existing systems, and implementation commitment. Contact AiBuildrs for a consultation regarding your specific situation.