AI is meaningfully automating several categories of legal work right now:

Document review and due diligence: AI-powered document review tools (Relativity, Kira, Luminance) can review thousands of contracts in hours — work that previously required teams of junior associates billing at $200–$400/hour. In M&A due diligence, AI is now standard. This has already reduced headcount in document-heavy practice areas and at document review firms.

Contract drafting and analysis: AI tools can draft standard commercial contracts, flag non-standard clauses, and compare contract terms against playbooks faster than manual drafting. For routine agreements — NDAs, service agreements, employment contracts — AI drafts are increasingly replacing associate first drafts.

Legal research: Tools like Westlaw AI, Lexis+ AI, and CoCounsel can conduct legal research, summarize case law, and identify relevant precedents faster than manual research. A task that took a junior associate 6 hours can now take 30 minutes. This is compressing the billing model for research-heavy work.

E-discovery: AI-assisted e-discovery has been automating document classification, privilege review, and relevance scoring for over a decade. This is the most mature area of legal AI and has already substantially reduced the headcount required for discovery-heavy litigation.

The legal work that remains most human-dependent:

Complex legal judgment in novel situations: Advising a client on how a court will likely rule on an unsettled question of law, structuring a transaction to achieve a specific tax treatment, or crafting litigation strategy for a case with genuinely novel facts requires the kind of contextual legal judgment that AI cannot reliably produce. AI can research precedent; it cannot predict how a specific judge will weigh competing considerations in a case it has never seen.

Client counseling and relationship: The attorney-client relationship — particularly in high-stakes situations (criminal defense, complex litigation, major transactions, family law) — depends on trust built through human relationship. Clients in difficult situations need counsel they trust, not an AI system. This is not sentiment; it is the mechanism through which effective legal representation works.

Adversarial advocacy: Oral argument, cross-examination, negotiation, and the read-the-room judgment required in courtroom and boardroom settings are deeply human activities. A skilled trial lawyer reading a witness, a deal lawyer sensing when a counterparty is bluffing, a mediator identifying the settlement zone — these require the kind of real-time human judgment and interpersonal intelligence that AI cannot replicate.

Ethics and professional judgment: Legal ethics — knowing when to withdraw, how to handle conflicts, what advice serves the client's genuine interests vs. their stated desires — requires the kind of moral reasoning and professional accountability that cannot be delegated to an AI system.

Not all legal careers face the same AI exposure:

Highest risk: Document review attorneys and contract review paralegals working on high-volume, routine work. Law firm associates whose billable hours are primarily driven by document review, research, and first-draft contract work — the exact tasks AI handles well. Legal process outsourcing (LPO) firms built around volume document work.

Moderate risk: General practice attorneys handling routine transactional work (residential real estate, basic estate planning, standard commercial agreements). Solo practitioners and small firm attorneys who compete primarily on price rather than specialized expertise.

Lowest risk: Trial lawyers and litigators with strong courtroom skills. Partners with deep client relationships and institutional trust. Attorneys in highly specialized practice areas requiring deep technical expertise (patent litigation, international arbitration, complex tax, criminal defense). General counsel and senior in-house counsel doing strategic legal work.

The legal professionals best positioned in the AI era:

Develop AI tool fluency immediately: Attorneys who can use Westlaw AI, Harvey, CoCounsel, and contract AI tools effectively are more productive than those who can't. This is currently a competitive advantage; it will become a baseline expectation. Law schools are beginning to require AI tool literacy.

Invest in the judgment and relationship skills AI cannot replicate: Trial advocacy, client counseling, negotiation, and the judgment to advise clients in genuinely complex situations are the skills that remain distinctly human and distinctly valuable. These are worth developing deliberately — through pro bono work, moot court, client-facing roles, and mentorship with experienced practitioners.

Specialize in AI-resilient practice areas: Criminal defense, complex litigation, family law in high-conflict cases, patent litigation, international arbitration, and regulatory work requiring deep agency expertise are all more AI-resilient than high-volume transactional commodity work.

For law students: Be honest about what you want to do in law. If your goal is document review work at a contract firm, AI exposure is real and near-term. If your goal is trial work, deal-making, or building a client practice, the AI transition creates opportunity — the attorneys who can combine AI productivity with human judgment will be the most valuable lawyers of the next decade.