
Every few years, a new wave of technology sweeps through the legal profession and a chorus of optimists predict the end of stenography. I’ve heard it for two decades in courtrooms from San Diego to Sacramento. First it was digital recorders. Then came automated speech recognition (ASR). Today the buzzword is “AI.”
The argument is always the same – machines are faster, cheaper, and soon will be good enough to replace trained reporters. But what I have seen in courtrooms across California tells a very different story. And what the data shows from Stanford and the National Court Reporters Association (NCRA) backs it up – replacing certified reporters with software is not just risky — it undermines accuracy, fairness, and public trust.
What the Data Says
In 2020, Stanford researchers (Koenecke et al.) tested five major ASR platforms from Amazon, Apple, Google, IBM, and Microsoft. The results were alarming: these systems made twice as many errors transcribing Black speakers as white speakers, with error rates averaging 35% compared to 19%. Those aren’t harmless typos. In a courtroom, “I did not do it” rendered as “I did do it” changes the trajectory of a case.
The NCRA, in its 2023 white paper “Emerging Ethical and Legal Issues Related to the Use of Artificial Intelligence (AI), Automatic Speech Recognition (ASR), Voice Cloning, and Digital Audio Recording of Legal Proceedings,” issued clear warnings. ASR fails most visibly in multi-speaker environments, with cross-talk, overlapping testimony, and technical terminology. The paper highlights risks including bias, misattribution of speakers, privacy concerns, and the absence of any clear chain of custody for digital records.
Taken together, the Stanford data and NCRA’s analysis confirm what reporters already know: ASR isn’t courtroom-ready for official records.
What I’ve Seen in Courtrooms
In one trial, a judge asked me to read back a disputed exchange after her courtroom ASR system produced something entirely different from what was actually said. My stenographic notes matched the attorneys’ memory word for word. The machine’s transcript didn’t. It was confident — and it was wrong.
In depositions, especially on Zoom, I’ve seen a new trend emerge. Attorneys sometimes run ASR software in the background, hoping to generate their own transcript. Reporters often catch them and stop it, but some slip through. In those cases, the reporter has been replaced by a videographer or a notary — neither authorized as a deposition officer. The Moment the Notary Loophole Was Unleashed in a Firestorm documents how many of these “reporter-free” depositions have proliferated under a legal interpretation that substitutes oath-administering notaries for stenographers, putting the admissibility and reliability of these transcripts at risk. The result is an uncertified record that’s vulnerable to hearsay objections and ethical challenges.
These are not minor glitches. They are structural failures that show why legal proceedings require certified human oversight.
What To Tell the AI Doomsayers
The legal world has no shortage of AI evangelists. They argue that reporters are expensive, that machines never call in sick, and that the public wants efficiency above all else.
Here is what to tell the AI doomsayers:
- Look at the data. Stanford proved racial bias persists in every major ASR system.
- Listen to the profession. The NCRA’s national white paper warns of accuracy, custody, and ethics failures.
- Watch the courtroom. Judges and attorneys turn to certified transcripts when the stakes are high.
Court reporters — whether machine or voice stenographers — are trained professionals who stop proceedings to clarify, who ask speakers to repeat, and who certify the transcript under penalty of perjury. No software program takes an oath.
The Hawk’s Eye
I tell my students that a reporter is like a hawk circling above the courtroom. We see every movement, every shift in tone, every layered interruption. A microphone, whether attached to an ASR engine or a digital recorder, only collects noise. Hawks hunt meaning. Microphones don’t.
That difference matters. In appeals, in cross-examination, in the record that outlives the trial itself, accuracy is not negotiable.
The Hearsay Problem
Under the rules of evidence, an uncertified transcript produced by ASR or digital recording is not admissible in court. It is hearsay: an out-of-court “statement” generated by a machine, with no human declarant to swear to its accuracy.
By contrast, a certified transcript produced by a licensed stenographer is not hearsay. The reporter is present in the proceeding, is an officer of the court, and signs under penalty of perjury that the transcript is a true and correct record. That certification converts the transcript from hearsay into admissible evidence.
I’ve seen this play out in practice. Attorneys have tried to use “raw” ASR output to impeach a witness. The judge rejected it outright: it was hearsay, unreliable, and unsigned. But when I produced the certified transcript, it was admitted without hesitation.
This is a distinction no software company can erase. Until AI can stand as a legal declarant — which it cannot — uncertified transcripts remain inadmissible.
The Legal Stakes
Consider how transcripts are used. They form the backbone of appellate review. They are quoted back to witnesses in deposition impeachment. They anchor motions for summary judgment. A mistranscribed word isn’t just an error — it can shift the course of litigation.
Judges know this. In California, I’ve had judges explicitly ask for my read-back after their ASR feed displayed something entirely different from what was spoken. They may be intrigued by the promise of automation, but when accuracy is questioned, they fall back on the human reporter. Every time.
Why It Matters
The debate is not “AI versus humans.” It is about standards. Trained stenographers, whether working on a machine or by voice, undergo years of study, thousands of practice hours, and licensing examinations. We are regulated professionals. We understand the difference between “I know” and “I no.”
ASR cannot yet deliver that level of contextual understanding. And until it can — without bias, without error, without loss of custody — the official record belongs in human hands.
The Bottom Line
Stanford’s numbers, NCRA’s warnings, and decades of courtroom experience converge on the same point: technology is a tool, not a replacement. Reporters generate realtime feeds and rough transcripts directly from their stenographic notes — not from machine output. Accuracy flows from the reporter’s own record, which is then refined and certified. That’s why attorneys and judges continue to rely on human reporters: the transcript originates with a trained officer of the court, not with a fallible algorithm. Judges may experiment with AI summaries. But the certified transcript, the one that carries the weight of law, still requires a reporter’s signature.
Until a machine can raise its right hand and swear to accuracy under penalty of perjury, court reporters remain indispensable.
That’s what to tell the AI doomsayers.
StenoImperium
Court Reporting. Unfiltered. Unafraid.
Disclaimer
“This article includes analysis and commentary based on observed events, public records, and legal statutes.”
The content of this post is intended for informational and discussion purposes only. All opinions expressed herein are those of the author and are based on publicly available information, industry standards, and good-faith concerns about nonprofit governance and professional ethics. No part of this article is intended to defame, accuse, or misrepresent any individual or organization. Readers are encouraged to verify facts independently and to engage constructively in dialogue about leadership, transparency, and accountability in the court reporting profession.
- The content on this blog represents the personal opinions, observations, and commentary of the author. It is intended for editorial and journalistic purposes and is protected under the First Amendment of the United States Constitution.
- Nothing here constitutes legal advice. Readers are encouraged to review the facts and form independent conclusions.
***To unsubscribe, just smash that UNSUBSCRIBE button below — yes, the one that’s universally glued to the bottom of every newsletter ever created. It’s basically the “Exit” sign of the email world. You can’t miss it. It looks like this (brace yourself for the excitement):

I don’t know who you are, Stenoimperium, but I thank you for ALL you are doing for our amazing profession! You are a LIGHT in the DARKNESS! I am sharing all of your articles with my employees and asking them to share with our clientele. I have been fighting the “good fight” on the freelancer side of the profession for 40 years. An idea for an article would be the stenographer’s “notes.” The stenographer’s notes never lie and are uncontroverted.
LikeLike
Thank you so much for these kind words — they mean more than you know. Forty years of fighting the good fight is no small thing, and I have immense respect for freelancers who’ve carried this profession forward. You’re absolutely right about the power of our notes: they are the one part of the process that’s unshakable, uncontroverted, and unimpeachable. That idea is already sparking thoughts for my next piece. Thank you for sharing the articles with your team and clients — that’s how our message gains traction. Together, we keep the light on in the darkness.
LikeLike
You asked, and I delivered:
LikeLike
Oh, did you deliver! What an EXCELLENT article. You just nailed it! Truly, I cannot thank you for ALL you are doing. I wish I had your eloquence. We will PERSEVERE!
LikeLike
Can you do me a little favor? Would you mind posting this under the article itself? TIA! I’m so glad it resonated with you — thank you, it truly means a lot. Your encouragement keeps me going. And yes… we will persevere, and we will win!
LikeLike