AI Replacing Lawyers: Tasks, Limits, and Human Judgment
AI assists lawyers with document review, research summaries, and drafting, but professional responsibility, client judgment, and regulated legal advice require a licensed attorney. Here is where the line is.
Updated 5 min read
AI is not replacing lawyers. It is replacing parts of the work that trained lawyers spend time on but that do not require a lawyer's judgment to do: first-pass document review, summarizing case law, drafting standard contract clauses, organizing deposition notes. The parts that require a lawyer remain with one: advising a client, applying legal judgment to specific facts, signing a filing, and taking professional responsibility for the outcome.
The distinction matters because it shapes how law firms and legal teams can responsibly use AI. Use it to speed up the tasks that do not require professional judgment. Apply human judgment to everything that does. Tools like fenbs help legal teams record the review rules and boundaries they have agreed on, so every AI-assisted workflow starts from the same standards, not from what an individual lawyer remembered to specify.
Document review and contract analysis
First-pass document review is one of the most time-consuming tasks in litigation and transactional work. AI tools can review large document sets for relevance, privilege, and specific terms faster than a team of first-year associates. Several e-discovery platforms have offered AI-assisted review for years; newer general-purpose AI tools can assist with contract review tasks such as identifying non-standard clauses, comparing against a playbook, or summarizing a document's key provisions.
The lawyer's role in document review does not go away with AI assistance; it shifts. A lawyer using AI for document review is responsible for defining what the AI looks for, reviewing the results for accuracy, and making the final privilege and relevance determinations. The AI is faster; the lawyer is accountable. Bar rules on supervision of non-lawyer assistance apply.
Research and precedent summarization
AI can summarize a body of case law, identify cases that support or challenge a position, and organize research into a memo structure faster than manual research. Several legal research platforms incorporate AI summaries into their existing tools. A lawyer using AI research assistance is responsible for verifying that the cases cited exist and say what the summary reports. AI hallucinations in legal research, where the AI cites a case that does not exist or misrepresents a holding, are a documented risk.
Verify every case citation against the primary source before filing or sending. This is not optional. An attorney who submits a brief with a fabricated citation faces sanctions under Rule 11 of the Federal Rules of Civil Procedure and bar discipline under Model Rules 3.3 and 8.4. The convenience of AI research does not reduce the duty to verify.
Drafting and first-pass editing
AI drafting tools can produce first drafts of contracts, demand letters, discovery requests, and motion sections from a brief and detailed instructions. The drafts need substantive review, not just proofing: an AI that produces a competent-looking contract clause may miss a jurisdiction-specific requirement or a business context it was not given. The reviewing lawyer checks both the legal accuracy and the fit with the client's actual situation.
AI can also improve existing drafts: flagging inconsistent defined terms, identifying passive voice in places where active is clearer, or suggesting a shorter version of a provision. These are editorial tasks. The lawyer decides whether the suggestions improve the document for this client and this matter.
Record your review rules where every AI session can read them
fenbs Decisions and rules holds the review requirements your team has agreed on: which outputs require a supervising attorney's sign-off, which document types need citation checking, and what confidentiality rules apply to client matter information. Every connected assistant reads those rules first. A new team member or a new AI tool starts from the same standards.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
Tasks that require a licensed attorney
Giving legal advice to a client requires a licensed attorney. Advising means applying legal analysis to a specific person's facts and recommending a course of action; providing information about the law is different. The unauthorized practice of law is a crime in every US jurisdiction. AI tools that provide legal information are not providing legal advice, but the line is not always obvious to a non-lawyer using the tool.
Representing a client in court, signing a pleading, or certifying to a tribunal requires an attorney. Negotiating on behalf of a client involves professional responsibility obligations that attach to the attorney. Taking a deposition, advising on litigation strategy, and counseling a client about risks are all within the practice of law. These do not change with AI assistance.
Professional responsibility and AI in law
Every US state bar association has ethics rules that apply to AI use in legal practice. The American Bar Association (ABA) has issued formal opinions and guidance on AI; several state bars have issued their own. The core requirements are consistent: competence in using the technology, supervision of AI output, confidentiality protection for client information, and honesty about AI use where required.
Confidentiality is the most sensitive issue. Client information entered into a general-purpose AI tool may be processed outside the attorney-client privilege and outside the firm's data governance. Review your firm's policies and your bar's guidance before entering any client matter information into an AI tool. Several courts have adopted standing orders requiring disclosure of AI use in filings; check the local rules for any jurisdiction where you practice.
Where review rules come from
The rules for AI use in a law practice come from multiple sources: state bar ethics rules, ABA guidance, court standing orders, firm policy, and client agreements. A legal team that keeps those rules in a single, findable place is better positioned than one that relies on each lawyer to remember the requirements for each matter. For the related question of AI replacing teachers, that page addresses a similar professional boundary in education. For AI agents for lawyers and AI agent task examples, those pages cover more of the operational picture.
How to do this with fenbs
Use fenbs to keep your legal AI workflow rules, review requirements, and lessons in one place.
- Create a free account and add a project for your legal AI workflow.
- Write the review rules in Decisions and rules: which output types require supervising attorney sign-off, which require citation verification, and what confidentiality rules apply to matter information.
- Add a task for each matter or matter type where AI assistance is in use. Record where the work stands and who is responsible for review.
- When AI assistance produces an issue, add a Lessons learned entry. Note what happened, what you checked, and what the review step should be next time.
- Connect your AI assistant from Integrations. It reads the rules at the start of each session so review requirements are explicit before work begins.