Five legal roles that arrive at the same structural worry from different directions: the routine work that trains junior lawyers is precisely the work being automated.
Sports Lawyer, Large Firm
Contract review compresses while a new docket of AI disputes grows.
A sports lawyer at a large firm is typically an associate or partner in a Sports & Gaming, Media & Entertainment, or Corporate practice group. The work centres on negotiating and drafting agreements (athlete endorsement and NIL deals, sponsorship, media rights, franchise sales), advising leagues, teams, and athletes on regulatory questions, and litigating disputes over eligibility, likeness rights, and league governance.
If AI erases that rung, the profession faces a long-term talent crisis.
Axios (2026) — the rung being the routine contract and diligence work that BigLaw has always used to teach junior associates judgment
How AI is used today
The AI stack is mostly inherited from the firm's general corporate and litigation tooling rather than built for sports: Harvey is used at more than half of AmLaw 100 firms for contract analysis and due diligence, and tools like Litera Kira and Luminance extract clauses — exclusivity windows, termination and clawback provisions — across the volume contract work. A genuinely sports-specific use is emerging in athlete-likeness verification, as firms are asked to assess AI-altered promotional videos and deepfaked recruiting footage; "athletes using AI to fabricate or enhance their own performances" is now a live disciplinary and legal issue in college sports.
What’s changing
The routine, templated side — NIL and sponsorship contract review, standard due diligence — keeps compressing, while a new docket of AI-caused disputes grows: deepfaked athlete likeness, AI-disclosure compliance for sportsbooks, market-integrity questions tied to prediction-market trading. But the more structural risk is not which tasks change. BigLaw's associate-training model depends on that same routine work to teach judgment, and "if AI erases that rung, the profession faces a long-term talent crisis." If the entry-level tasks that trained junior lawyers are automated, firms may have to redesign how associates build judgment at all rather than simply reassign their hours.
What a new entrant needs
Beyond baseline tool fluency, the most important skill is judgment about when output can and cannot be trusted — courts have already sanctioned attorneys for filing briefs with AI-fabricated citations. That is Ethical and Responsible Use: knowing what client or case information is safe to enter into a tool before verifying the platform's confidentiality and data-handling terms. The risk is not hypothetical; even at in-house teams furthest along on adoption, "concerns about data privacy and confidentiality (59%)" remain among the most cited barriers. A junior lawyer has less firm-specific experience to judge that risk, and law school — unlike practice — does not impose client-confidentiality rules, which is why completing a firm's AI and confidentiality training before independent use matters more for a new associate than for anyone else in the building.
Where the evidence disagrees
Headline employment numbers need reading carefully. Goldman Sachs' 2025 workforce update frames its figure conditionally: if current AI use cases were expanded economy-wide and reduced employment proportionally to efficiency gains, roughly 2.5% of overall U.S. employment would be at risk, with legal and administrative support among the more exposed — while the report's own authors "remain skeptical that AI will lead to large employment reductions" over the next decade. That is a hypothetical scenario, not a measured outcome.
4 sources, located and verified by the contributor
Put an unfiled invention into the wrong tool and you can destroy the patent.
Research by Ava Desanti
Typical employers IP boutique firms · IP groups within general-practice firms · in-house legal departments at technology and pharmaceutical companies
What the job is
Patent prosecution attorneys draft, file, and shepherd patent applications through examination before the USPTO, translating an inventor's technical disclosure into legally defensible claims and then negotiating those claims through examination.
Simply relying on the accuracy of an AI tool is not a reasonable inquiry.
USPTO guidance (2024)
How AI is used today
AI is already used on both sides of that process. Examiners use AI-enabled prior-art search built into the USPTO's internal system, including More Like This Document and Similarity Search. On the practitioner side, attorneys increasingly use generative AI to draft technical specifications, prepare responses to office actions, and draft claim language. The USPTO has been explicit that no output can simply be trusted: its 2024 guidance states that "simply relying on the accuracy of an AI tool is not a reasonable inquiry," and every AI-assisted filing must be independently reviewed and verified by the signing attorney. The Office itself is still testing — in Q4 of fiscal 2025 "the USPTO deployed a USPTO-approved generative AI tool to supervisory patent examiners" to evaluate quality and time savings before any broader rollout.
What’s changing
Two changes are likely within five years, with sharply different confidence. The first is close to certain: the USPTO is actively building AI tools to help examiners with claim construction, Section 101 eligibility analysis, and invention summaries, with guardrails stakeholders describe as non-negotiable — including that "examiners remain the authors of record and final decision-makers." The second is far less settled: patent eligibility law itself is split between a USPTO that has become notably more permissive toward AI-related inventions and a Federal Circuit that, as one panellist put it, has said "We're not gonna follow your framework." A patent attorney may need to draft claims that survive two diverging legal standards at once, which makes it genuinely hard to predict what normal prosecution looks like five years out.
What a new entrant needs
Beyond drafting skill, a new patent attorney needs disciplined judgment about what can safely go into an AI tool. Summarizing already-public prior art is low risk. Entering a client's unfiled technical invention details into a consumer-grade tool is not, because free-tier providers' terms "often reserve the right to store your prompts and use them to train and refine the model" — and the information itself can become a public disclosure that destroys patent rights immediately. This is Ethical and Responsible Use at the Fluency level: applying institutional and professional guidelines in live client matters rather than understanding them abstractly. The stakes are concrete and often irreversible, which is why the discipline here is simple and strict: file first, disclose later.
Where the evidence disagrees
A December 2025 Berkeley Law patent panel transcript attributes the Ex parte Desjardins decision to two different USPTO directors within the same conversation — one speaker says "Director Squires," another says "Director Kathi Vidal." Cross-referencing a second independent source supported Squires as the more likely attribution, but the inconsistency exists in the primary source itself and is flagged here rather than silently resolved.
4 sources, located and verified by the contributor
Code-parsing automates. The hearing where a board exercises discretion does not.
Research by Duncan Keefe
Typical employers Boutique land use and zoning firms · real estate developers with in-house counsel · environmental law firms · municipal law departments
What the job is
A land use attorney represents developers, municipalities, and residents in disputes over how a property may be used — work that plays out before local planning boards, zoning boards of appeal, and conservation commissions more often than in a courtroom. The job involves drafting and defending applications for special permits, variances, and site plan approval, negotiating conditions with abutters and regulators, and filing suit when the administrative process fails. Zoning analysis has traditionally meant a lawyer reading a municipal code of hundreds of pages, matching a proposed use against permitted-use tables scattered across base zoning, overlay districts, and special area plans.
[An AI tool is already] capable of reproducing the syntax of a zoning ordinance.
Morley (2023), on an AI tool tested against municipal code drafting
How AI is used today
Firms now use tools built to automate exactly that parsing — one firm's own analysis describes the technology as aimed at "automating the analysis of complex codes, streamlining compliance checks, and providing real-time insights for legal professionals and planners," which is precisely the groundwork that used to fall to junior associates. On the environmental review side, federal agencies have begun piloting AI in drafting Environmental Impact Statements under NEPA, with the Department of Energy developing a platform, PermitAI.
What’s changing
Within five years the paperwork side — matching a parcel to its zoning designation, calculating setbacks and floor area ratios, pulling relevant provisions out of a thousand-page code — is likely substantially automated, since those are structured lookup and extraction tasks. One planning experiment found an AI tool already "capable of reproducing the syntax of a zoning ordinance," which suggests the drafting side of local government work changes nearly as much as private practice. What is less likely to change is the discretionary, adversarial part: the hearing where a board decides whether a variance causes substantial harm to the public good, or where an EIS is challenged in court over whether the agency gave the evidence a hard look. Judicial review is already struggling with "the opacity, bias, and limited explainability of AI systems" in exactly these contested cases, and that tension — AI's speed on routine compliance against its unreliability on contested judgment — is more likely to define the practice area than to resolve.
What a new entrant needs
A junior land use attorney needs the fluency to use AI for the compliance-heavy lifting, but the dimension that matters most is Critical Thinking and Judgement — specifically the expectation that a lawyer cross-references AI output against authoritative sources before relying on it. That is not hypothetical here. A zoning report or an EIS built on a hallucinated or outdated code section is a document opposing counsel will find and use against you, and a junior associate who cannot tell a real ordinance provision from a plausible-sounding one is a liability rather than an asset.
Where the evidence disagrees
Firms advising developers are telling clients that any AI-assisted agency review must be "properly supervised, and follows recognized best practices," because courts have not settled how much AI-generated analysis a federal environmental review can lean on before it violates the hard-look standard under the Administrative Procedure Act.
5 sources, located and verified by the contributor
Boards are using AI far faster than they are governing it — and that gap is the job.
Research by Dylan Aufman
Typical employers Large and mid-size consumer-goods companies
What the job is
A general counsel is a company's top in-house lawyer, overseeing regulatory compliance, drafting major contracts, and managing liability exposure and risk — often supervising a department of subordinate lawyers.
Boards are using AI faster than they are governing itFigures are percentages. Both figures are from the same 2026 survey of directors. Data: Diligent (2026). Chart drawn for this page; the underlying figures are the contributor’s verified sources.
How AI is used today
AI has entered the day-to-day through contract work. Spellbook, an AI contract-review tool built as a Microsoft Word extension rather than a standalone platform, launched in 2022 and by October 2025 had a customer base of 4,000 across 80 countries, aimed at in-house departments and firms handling large contract volumes. Clause analysis has followed: "document review is now the second most common GenAI use case among legal professionals (74%), trailing only legal research (80%)." That saves hours of bulk review and, for a general counsel who has historically delegated the work, frees capacity to delegate further up the value chain.
That gap is the job. A general counsel is already the overseer of compliance and consistency, which makes them the obvious owner of company-wide AI governance.
What’s changing
One plausible direction is that general counsels take on governance of how their entire company uses AI — writing company-wide guidance and creating consistency across departments. A 2026 study found 66% of directors reported using AI for board work while only 22% reported having formal AI governance processes in place. As the key overseer of regulatory compliance and departmental consistency, a general counsel is well positioned to manage that gap, particularly if governments impose stricter AI regulation. There is reasonable scepticism about how quickly this happens.
What a new entrant needs
Some level of AI fluency is becoming vital, and ethical use is the specific pressure point: a general counsel may have to police unethical or even illegal AI use inside their own company. Doing that requires strong critical thinking and a deep working knowledge of the intersections between AI, morality, and law — which adds to rather than replaces the leadership dimension of the role.
3 sources, located and verified by the contributor
AI is automating the research and generating a new category of case at once.
Research by Nandini Shah
Typical employers Amnesty International · Human Rights Watch · refugee-rights groups · nonprofits
What the job is
Human rights lawyers at NGOs provide legal representation to individuals or groups whose human rights have been violated.
AI is changing this role from both ends at once — automating the research and drafting that fills a junior lawyer’s week, while generating an entirely new category of case to bring.
How AI is used today
AI shows up in three places: legal research, where tools like CoCounsel search case law, summarize documents, and identify relevant authorities; document analysis, summarizing reports, testimony, and evidence; and drafting, generating initial versions of briefs, complaints, and memos. The aspects most exposed to automation are routine research, document review, summarization, and first drafts.
What’s changing
Less time on manual research and summarization means more time on legal judgment, strategy, and advocacy. At the same time the substance of the work is shifting, because AI is generating its own human rights docket — privacy, discrimination, deepfakes. A third change is harder: automated decision-making is entering courts, and lawyers will need to evaluate whether that is ethical. The direction is clear; the pace and extent are uncertain given how fast both the technology and the regulation are moving.
What a new entrant needs
Practical AI skills for research, summarization, and drafting; strong judgment and independent verification of outputs against accurate sources; and privacy discipline, because witness testimony and confidential case documents are exactly the material that must not leak into a tool — which is the Ethical and Responsible Use dimension. Beyond using AI, a new entrant should understand AI as a human rights issue, since a growing share of cases involve AI-related risks and harms.
5 sources, located and verified by the contributor
Mashayekhi, M., & Nosrati, F. (2025). Human-AI collaboration in legal services: Empirical insights on task-technology fit and generative AI adoption by legal professionals. SAM Advanced Management Journal, 90(3), 305–333.
Nielsen, A., Skylaki, S., Norkute, M., & Stremitzer, A. (2024). Building a better lawyer: Experimental evidence that artificial intelligence can increase legal work efficiency. Journal of Empirical Legal Studies, 21(4).