The Court Reporter Shortage Is Getting Worse — and Legal AI Is Both the Problem and the Only Credible Solution

The United States is running out of court reporters, and the legal profession has spent the better part of a decade treating this as a calendar problem. It is not. It is an infrastructure crisis, and the gap between what the litigation system demands and what it can actually deliver is widening fast.

The Numbers Are Not Ambiguous

The National Court Reporters Association (NCRA) has been flagging a demographic cliff for years. The average age of a certified court reporter in the U.S. is somewhere north of 50. Stenography programs have been closing — enrollment dropped sharply through the 2010s as prospective students looked at three-to-five years of rigorous training for a profession that seemed perpetually on the verge of automation. The result is that attrition is dramatically outpacing credentialing. The field will likely need to replace a large share of its workforce over the next decade while simultaneously meeting increased demand.

In Texas, deposition scheduling delays in Dallas and Houston metros can stretch to several weeks for routine civil matters. Florida, including the Miami-Dade and Broward markets, may face similar pressure on certified reporter availability for complex commercial cases. The Southeast broadly, from Atlanta to Charlotte, is seeing similar compression. These are not aberrations. They are the new baseline.

Remote Tooling Papered Over a Structural Problem

When the pandemic forced depositions online in 2020, the profession largely congratulated itself on its adaptability. Platforms like Zoom, Veritext's virtual deposition suites, and Planet Depos remote services allowed litigation to continue. Court reporters dialed in from home. The optics were fine.

What actually happened is that remote capability disguised the shortage by drawing on a geographically distributed pool of reporters. A deposition in Austin could be covered by a certified reporter in Ohio. This worked until the national pool itself became the constraint. You cannot route around a shortfall that is systemic. Remote deposition infrastructure solved the geographic problem and exposed the headcount problem.

Legal ops leads at large firms noticed this around 2023 and 2024, when even remote bookings started hitting multi-week delays. The scheduling problem had simply moved upstream.

AI-Assisted Litigation Is Making This Worse, Not Better

Here is the uncomfortable part: legal AI tools are exacerbating the shortage they are supposedly positioned to solve.

AI-assisted discovery, contract review, and litigation strategy platforms — think Harvey, Casetext's CoCounsel before it was absorbed into Thomson Reuters, and the expanding suite of litigation AI tools from major providers — have materially lowered the cost of identifying deposition targets. When it is faster and cheaper to map out a deposition strategy across twenty witnesses, attorneys take more depositions. The volume pressure on court reporter supply has increased in direct proportion to how efficient AI has made pre-deposition preparation.

Litigation boutiques using AI-assisted discovery workflows can plausibly take noticeably more depositions on comparable case loads than they did five years ago. More depositions require more reporters. The math is straightforward and unflattering.

What AI Transcription Can and Cannot Do

The obvious response is AI transcription, and several vendors are now offering automated deposition transcription services at a fraction of the cost of a certified reporter. Otter.ai, Verbit, and specialized legal transcription platforms like Steno's AI offerings have entered this space aggressively.

The accuracy question is real but often overstated in the wrong direction. For clean audio with standard American English speakers, current AI transcription accuracy is genuinely impressive — often very high on word-level accuracy. The problem is that depositions are not clean audio environments. Accented speech, crosstalk, technical terminology, and the specific cadences of deposition colloquy all degrade performance meaningfully. A 3 percent error rate on a four-hour deposition transcript produces a lot of errors.

The admissibility question is more serious. Federal courts and most state courts require that deposition transcripts be certified by a qualified reporter. The Federal Rules of Civil Procedure, specifically Rule 28, require depositions to be taken before an officer authorized to administer oaths who is not a party. Many jurisdictions do not appear to have updated their rules to contemplate AI-generated transcripts as a primary record, and there is little case law on the question. Courts can be expected to scrutinize transcripts that cannot trace to a certified human officer. AI transcripts used without a certified reporter present may also raise chain-of-custody questions.

What Courts and Bar Associations Are Actually Doing

Mostly not enough, and mostly too slowly. A handful of states have begun exploring certified legal transcriptionist (CLT) programs as a lower-barrier credential that could partially substitute for full stenographers in some deposition contexts. The NCRA has been resistant to this, understandably given member interests, but the organizational resistance has slowed what should be a policy conversation.

A sensible starting point would be allowing AI-assisted transcription in limited ancillary proceedings where the transcript is not being offered as evidence. This is a start. It is also approximately five years behind where the technology and the market pressure already are.

This Is an Infrastructure Problem, Not a Scheduling Problem

The legal profession has a documented tendency to treat systemic operational failures as vendor management issues. The e-discovery cost crisis spent fifteen years being handled as a matter of negotiating with Relativity before anyone treated it as a process redesign problem. The court reporter shortage is following the same pattern.

Legal ops leads who are serious about this should be doing three things now: building preferred vendor relationships with regional court reporting agencies while that still buys something, piloting AI-assisted transcription in non-evidentiary contexts to develop internal quality benchmarks, and engaging directly with state bar rule committees on updated certification frameworks.

The shortage is not going to resolve through market dynamics alone. The training pipeline is broken, the demographics are unfavorable, and AI deposition tooling is increasing demand faster than any credentialing program can close the gap. Treating this as anything less than a litigation infrastructure crisis is not caution. It is negligence dressed up as patience.