By Dorna Moini, CEO and founder of Gavel
The U.S. legal profession has grown from 114,000 lawyers in 1900 to 1.37 million today, according to the American Bar Association. That growth happened across six major waves of technology that were each predicted to shrink the profession. Every single wave expanded it instead. As someone who practiced law for many years and now builds legal technology used by thousands of organizations around the world at Gavel, I think this history is the most useful lens we have for understanding what AI will actually do to legal jobs. The ultimate question is whether AI is different than these other legal tech revolutions given that it, given that it touches the highest layer of knowledge work.
The Typewriter (1880s–1950s)
Before the typewriter, law firms employed rooms of scriveners whose sole job was copying legal documents by hand. One typewriter operator could produce what six clerks did with a pen, and the expectation was that these clerical roles would collapse. Instead, the occupation of typists and secretaries grew sixfold between 1900 and 1950, from under 0.5% of all U.S. employment to 3%. Firms could suddenly produce contracts, briefs, and correspondence at a fraction of the prior cost, so they produced dramatically more of it. The typewriter also opened professional employment to women at scale for the first time, with typing jobs paying as much or more than teaching, which had been one of the only professional paths available to women at the time.
The Word Processor (1970s–1980s)
The first word processing program appeared in 1976, and by the end of that decade, most law firms had adopted one. The fear was that legal secretaries would be wiped out. What actually happened is that the number of lawyers in America jumped 76% in the 1970s alone, from 326,000 to 574,000, according to the ABA. Word processors made it faster to draft, revise, and finalize documents, enabling firms to take on more matters per attorney. The secretary’s role evolved from retyping entire documents to managing workflows, coordinating filings, handling client communications, and providing office and administrative support, and office and administrative support employment peaked at 12.7% of all jobs in 1980, reflecting the expansion of the overall volume of professional services alongside these tools.
Computerized Legal Research (1970s–2000s)
Lexis launched in the early 1970s as the first commercial electronic database of case law, followed shortly by Westlaw. Both started as dial-up services with dedicated terminals. Before these platforms, legal research meant physically walking to a law library, pulling volumes off shelves, and spending hours tracing citations through printed indexes. A research task that used to take a junior associate an entire day could suddenly be done in an hour. The concern was that firms would need fewer associates. Instead, the profession grew from roughly 574,000 lawyers in 1980 to over a million by 2000. Faster research made legal services accessible to a broader client base, which expanded the overall market for legal work rather than compressing it.
E-Discovery Software (Late 1990s–2010s)
This is the example I find most instructive, because I lived through it. When electronic discovery software started replacing manual document review, the prediction was that paralegals and contract attorneys would be decimated. Technology-assisted review could process documents at a fraction of the cost of human review. But as economist James Bessen has documented, when legal offices adopted electronic discovery software in the late 1990s, the number of paralegals actually increased. The e-discovery industry itself grew to over $2 billion by 2021, spawning an entirely new job category in the process. E-discovery specialists now command average salaries above $80,000, and the Bureau of Labor Statistics projects paralegal employment to grow 14% from 2022 to 2032, well above the national average.
Document Automation (2010s–Present)
Tools that generate standard contracts, lease agreements, and incorporation documents from templates were supposed to eliminate junior associate work. And they did eliminate that specific task. But they also created new categories of work: building and maintaining those templates, customizing automation workflows for specific practice areas, quality-checking outputs, and, crucially, serving clients who previously couldn’t afford any legal help at all. The legal market grew from $713 billion in 2021 to $789 billion in 2022, and venture capital investment in legal tech startups jumped from $1 billion in 2018 to $4 billion by 2021. By 2025, projections suggested that half of all law firm employees would be non-lawyers with skills in technology, data analytics, and project management.
The Pattern Outside of Legal
This isn’t a legal-specific phenomenon. The most well-known case is the ATM. When automated teller machines rolled out starting in the 1970s, everyone assumed bank tellers were finished. By the mid-1990s, over 400,000 ATMs were installed across the United States, and the average branch went from needing 20 tellers to 13. But the reduced cost per branch meant banks opened 43% more branches in urban areas, and bank teller employment grew from roughly 300,000 in 1970 to 600,000 by 2010, according to James Bessen’s research for the IMF. Tellers shifted from cash handling to relationship banking and selling financial products. When barcode scanning technology hit retail, the number of cashiers also increased, because stores could process more transactions faster and expand operations to match.
What This Means for AI in Legal
I think about this constantly because it’s my job to. At Gavel, we build software that automates document generation by using rules-based logic (Gavel Workflows). We also have a widely-used AI assistant in Microsoft Word, Gavel Exec, which automates contract review, runs playbooks on contracts, and AI redlining. So we are very familiar with the entire spectrum of automation.
I’m not naïve about what AI can do. It can draft contracts, summarize case law, analyze discovery documents, and predict case outcomes. It will absolutely replace certain tasks that junior lawyers and paralegals do today. But if 140 years of legal technology history tells us anything, it’s that making legal work cheaper and faster doesn’t produce less of it. It produces more because more clients can afford it, more problems become worth solving, and more complexity still requires human judgment. The question is whether the gap in access to legal services will be enough to fill the automation of certain work.
The new roles I expect to see emerge include legal prompt engineers who specialize in getting the best outputs from AI tools for specific practice areas, AI audit attorneys who verify and quality-check AI-generated work before it reaches clients or courts, legal technology consultants who help firms select and optimize AI tools, access-to-justice specialists who use AI platforms to serve the roughly 80% of Americans who currently can’t afford a lawyer, and workflow architects who design the systems that combine AI, human review, and client interaction into something that actually works.
The lawyers and legal professionals who learn to work with AI rather than compete against it will win, and they will find themselves doing more interesting, more impactful, and higher-value work than the tasks these tools are replacing.
Dorna Moini is the CEO and founder of Gavel (www.gavel.io), the complete automation suite for lawyers. Prior to starting Gavel, Dorna was an attorney at Sidley Austin. Dorna is on the Legal Services Corporation Emerging Leaders Council and a member of LAFLA’s Advisory Board. She was named an ABA Legal Rebel and a Fastcase 50 honoree. She also teaches the Legal Innovations Lab at USC Law School.
The post Guest post: Six tech revolutions that were supposed to shrink legal but grew it. Will AI be different? appeared first on Legal IT Insider.