Law firms and companies' legal teams are experimenting with AI tools that can handle work done by entry-level lawyers, potentially reducing billable hours
Lawyers start to use GPT-4 technology to do legal research, draft documents and analyze contracts — Big law firms … LinkedIn: Elizabeth Smith , Damien Riehl , and Dazza Greenwood Tweets: @tprstly , @ajtoce , @adamsinger , @pt , @petehotchkiss , @wsj , @yoda , @ashleygjovik , and @casetext LinkedIn: Elizabeth Smith : As a survivor of the law firm billable hour, my heart is warmed each time an article is published suggesting that the billable hour is on its way to obsolescence. … Damien Riehl : Fodder for our profession's consideration. In the past two weeks, I've spoken with several world- class firms about this very topic. … Dazza Greenwood : It's hard to know when the billable hour will end as the main invoice and payment unit for lawyers but it's clear a lot fewer hours are now needed for a lot of legal work. … Tweets: Theo / @tprstly : *potentially removing jobs. Fixed it for you. https://twitter.com/... @ajtoce : Just like advisors, lawyers will say that each person's situation is different and requires personalized advice. Big firms and solos good at marketing will prevail. Those who are great at practicing law but bad at marketing will suffer. https://twitter.com/... Adam Singer / @adamsinger : Boomers hosed millennials on home/healthcare/edu costs; putting everyone into intractable debt and keeping positions of power forever. Millennials will hose zoomers by simply automating their work before they can get experience. Zoomers just gonna flip the table over... https://twitter.com/... Parker / @pt : One way to think about LLMs is they let us cheaply build text-based APIs to humans, which is all many lawyers are doing (eg contracts, filing forms, etc). Startups have been struggling to figure this out for a long time, but it feels like the productivity gains are finally here. https://twitter.com/... Pete Hotchkiss / @petehotchkiss : But how are entry-level layers meant to move on from being entry-level if they don't learn the ropes doing entry-level stuff? https://twitter.com/... @wsj : Big law firms are experimenting with artificial intelligence tools using an advanced version of ChatGPT that can handle the drudgery of drafting documents and reading contracts https://www.wsj.com/... Drew Olanoff / @yoda : Y'all gonna retract all this in 6 months or nah https://twitter.com/... @ashleygjovik : First: getting really tired of these oppressive headlines trying to drive fear in workers Second: has anyone explained ‘servers’ to these lawyers (re: work product & attorney-client privilege) Third: incoming “AI is practicing law without a license” complaints https://twitter.com/... @casetext : Today's @WSJ explores how #CoCounsel is reshaping legal practice. Our Fortune 50 partners Microsoft & Ford Motor—and top firms DLA Piper, Kirkland & Ellis, Skadden, and Orrick—played key roles as beta testers. Read about CoCounsel's progress since then. https://www.wsj.com/...
Context & Ripple Effects
The automation of junior-lawyer work has a long runway: a 2017 Financial Times look at Ravn, Luminance, and Lex Machina documented startups already absorbing tasks done by juniors and paralegals. What changed by spring 2023 was generative reach — Allen & Overy, among the world's largest firms, put OpenAI-backed Harvey to work writing legal documents, and GPT-4-based tools like Casetext's CoCounsel moved from novelty to production, with Microsoft, Ford, and firms including DLA Piper, Kirkland & Ellis, Skadden, and Orrick serving as beta testers.
First-order effects
- Entry-level lawyers at big firms and corporate legal teams see research, drafting, and contract analysis — their traditional workload — performed by GPT-4 tools, directly compressing the billable hours that fund the associate pyramid.
Second-order effects
- If AI absorbs associate-level output faster than firms cut headcount, the billable-hour model itself comes under pricing pressure, pushing firms toward fixed or value-based fees for work clients can now benchmark against machine output.
Third-order effects
- The profession's training pipeline is at stake: if the routine tasks that teach juniors judgment are automated, firms must redesign how associates learn, while adoption discipline matters — a New York lawyer's sanctions case over fabricated ChatGPT citations shows the failure mode, and US judges are already testing generative AI for research and drafting orders themselves.
The trend: Legal services are migrating from selling lawyer-hours to selling verified outcomes, with generative AI absorbing the entry-level work that anchored both firm economics and professional training.