Login | April 17, 2024
Court reporters look askance at AI-based transcripts
RICHARD WEINER
Technology for Lawyers
Published: June 24, 2022
There is a wave of new, artificial intelligence-based technology that threatens to overtake court reporters, said Susan Terry, past president of the Ohio Court Reporters Association (OCRA).
For the first time, outside parties to the court reporting profession are starting to move into territory that has traditionally belonged to a select group of individuals, and Terry and the rest of the profession is concerned.
“There has been a lot of turmoil in the industry in the last year or two because there are a lot of outside investors trying to jump into the business using AI transcription,” said Terry.
AI, of course, is far from the only technological disruption that the court reporting profession has seen in its several-hundred-year history.
Court reporting stenographic methods have undergone changes in technology virtually throughout its entire history.
According to Brookscourtreporting.com, taking notes on court processes is almost as old as writing itself, and shorthand systems date back at least to the time of the Roman Empire.
One of the first shorthand systems in English, capturing sounds rather than letters, was the 1837 tome Stenographic Sound-Hand by Sir Isaac Pittman. Pittman’s brother Benn brought the system to the US.
The first shorthand recording machine was invented by Miles Bartholomew in 1877, and the first modern, commercial stenotype machine was invented in 1906 by Ward Stone Ireland.
The steno machine that court reporters use today was developed over time out of this latter machine.
For many years, the machine’s paper tape, created by typing the stenographer’s code onto it, was transcribed by hand, with the court reporter sitting at a typewriter and translating that code into words.
Then, in the 1970’s, came Computer Aided Transcription (CAT), which translated the court reporter’s code into a real-time transcript. But CAT still needed (and still needs) human intervention for accuracy, especially in highly technical transcriptions like a medical deposition. Nevertheless, she said, certified court reporters who use CAT claim a 99 percent accuracy.
And accuracy is the main point, and the primary arterial clog, in discussing AI-assisted transcription, said Terry.
“AI reporting firms just set up a ‘digital reporter,’” she said, which is just a tape recorder. Some send out recorders with an affiliated software system.
The largest AI-based court reporting company is Verbit. They never returned several requests for comment for this article.
That recording is then fed into an AI program. And here is where the problems begin, she said. Court reporters need to be completely accurate, and AI court reporting is far from accurate, she said.
“No technology can do what we can do within the time and specifications,” said Terry.
There are several companies that try, she said.
According to her, these companies hire people who are not rained court reporters to run AI-based programs that transcribe recorded testimony. But the recordings themselves are often unclear, and the untrained transcribers may not know their way around the various legal rules that a certified court reporter needs to know.
“The testimony goes through multiple AI programs and is transcribed by multiple people who may or may not be in this country,” she said.
Terry feels that the growing use of AI-based transcription is not only eroding the work possibilities for court reporters, but is also producing inferior product. She also said that it can take a much longer time to produce a transcript using AI-based programs, and they may never be as accurate as real-time court reporters and their professional transcription techniques.
In fact, she said that courts have rejected AI-based transcripts over inaccuracies.
The American Bar Association has similar difficulties with the use of AI in the profession, which have culminated in 2019’s ABA House of Delegates Resolution 112.
This resolution “urges the courts and lawyers to address the emerging legal and ethical issues related to the use of artificial intelligence….”
The resolution talks a bit about ethical issues, but not about accuracy in transcription. And it does not offer any concrete solutions to the potential problems with AI that go well beyond the transcription problems.
Lots of smoke; no fire.
Terry said that AI transcription companies are on the verge of creating a labor crisis in the profession, as AI transcriptionists have far less training and far lower salaries than certified court reporters.
So the fight is apparently on: Court reporters vs. AI-based transcriptionists. We shall see.
But of course there is more on the court reporting technology front. Next on the tech horizon might be biometrics in court reporting, including voice identification, said Terry.
This might be especially useful as the use of remote depositions expands.