Will AI replace lawyers?
Unlikely in the near term. BLS projects lawyer jobs to grow 5% from 2025 to 2035 and says automating routine legal work isn’t expected to reduce overall demand. AI exposure is moderate: 138th of 756 occupations in Anthropic’s usage data. Research, interpretation and drafting are the most exposed tasks; representing clients in court and negotiating are the least. The early effect is on junior work, though AI’s role in slower associate hiring isn’t established.
- US jobs (2025)
- 863,700
- Projected change, 2025–35
- +5%
- Median pay (2025)
- $159,670
What does the research say?
- Microsoft Research (2025)AI applicability score: 0.180, ranked 286 of 785 occupations0–1 score from 2024 Bing Copilot conversations: how much of the occupation’s work people used AI for, how often it succeeded, and how broadly. Higher means more applicable.Source ↗
- Anthropic Economic Index (2026)Observed exposure: 0.17, ranked 138 of 756 occupations0–1 share of the occupation’s time-weighted tasks that AI could theoretically speed up and that already show up in work-related Claude usage. Higher means more exposed.Source ↗
Which parts of the job are most exposed?
- Advise clients concerning business transactions, claim liability, advisability of prosecuting or defending lawsuits, or legal rights and obligations ↗
- Interpret laws, rulings and regulations for individuals and businesses ↗
- Prepare legal briefs and opinions, and file appeals in state and federal courts of appeal ↗
- Represent clients in court or before government agencies ↗
What’s actually happening at companies?
- Aug 6, 2026NALP data: firms with 500+ lawyers hired 7.5% fewer entry-level associates from the Class of 2025 than a year earlier (about 540 fewer), the first decline since 2014. Article says AI's role is raised by observers but 'far too early to call' a trend.Reported ↗
- Jul 20, 2026SurePoint Law School Hiring Report 2026: new-associate hiring flat (7,489 in 2022; 7,426 in 2025). Artificial Lawyer suggests AI 'is playing into these figures' but says the size of the effect is unknown.Reported ↗
What happens when AI meets a merger?
In-house legal teams consolidate after an acquisition like any corporate function: two general counsels, two contract teams, two sets of outside firms. Deal work also creates a surge of contract review and diligence, exactly the tasks AI speeds up most, which then falls away once the integration ends. In the Acquisition Log, buyers from Broadcom (VMware) to Capital One (Discover) cut corporate roles anywhere from days to months after closing.
Every acquisition ends with a spreadsheet: roles, costs, overlaps. AI adds a column. As buyers get more data on what every employee does, and tools that can do part of it, the question stops being “do we need two of these?” and becomes “how many do we need at all?” That’s why the moves below matter more in a deal year than any other.
What should lawyers do now?
- Lean into the work AI can’t sign: advice, negotiation and representing clients.
- Use AI for first drafts and research, and be known for the judgment you add on top.
- If you’re junior, get in front of clients and partners early; that’s the work that can’t be automated.
- In-house during a deal? Volunteer for the integration: contract harmonization and entity cleanup last for months.
- Track matters won, risk avoided and dollars saved.
Questions people ask
Will AI replace lawyers?
Not according to current projections: BLS expects lawyer jobs to grow 5% from 2025 to 2035. AI is most useful for research and drafting; representing clients, advising on judgment calls and negotiating are the least exposed parts of the job.
Is AI reducing hiring at law firms?
Entry-level hiring has softened: firms with 500 or more lawyers hired 7.5% fewer first-year associates from the class of 2025, per NALP data reported by Above the Law. Whether AI is the cause hasn’t been established.
Find out where you stand before the org chart changes.
Take the Survival Fitness Score →Figures come from the U.S. Bureau of Labor Statistics and the published research linked above. Exposure measures estimate how much of an occupation’s work AI could help with or perform; they are not predictions of job losses.