With the rise of AI (Artificial Intelligence) we ask ourselves as transcriptionists, “Is this the technology that finally makes my career obsolete?” It’s a very legitimate question. Since I began transcribing in the Fall of 1990 I have continually heard, “One day soon technology will make your job obsolete.” I’ve always taken those assertions with a healthy dose of skepticism, and maybe some degree of denial, because who wants to believe their chosen field will become obsolete? In fairly recent history there are communication professions that are now unheard of such as telegraph operators, telephone operators, and with the exception of court reporters stenographers are now mostly passé. The old-time in-office typing pool with its cloud of cigarette smoke is now individuals scattered across the globe typing from the comfort of their own homes.
I went into medical transcription typing on an IBM Selectric typewriter (you could type a whole line appearing on a small screen that gave you a chance to edit the line before hitting “return” and then the machine printed the line). The trend to the electronic medical record was in it’s infancy, and the more rural hospitals, including the one where I worked, had not gone there yet. The month before I left that job they began the process of converting to computers for transcription. My next job was completely done on computer though we still printed the reports on paper. I had classes in computer technology in college and was young, so no biggie for me. Some of my older colleagues had a bigger adjustment, but they eventually caught on.
In the late nineties when I worked for a transcription company from home I transcribed via Dictaphone technology with a Dictaphone attached to a second line in my home. I dialed into the host computer and transcribed dictated reports into M.S. Word with my headset attached to the Dictaphone via a live phone connection. Once or twice a day (or more often if I had “stat” jobs) I would connect to the internet via dial-up and transfer the Word Docs back to the transcription company for whom I was working. I got my job lists once or twice per day via fax. Just typing that makes me feel about 90 years old, but that was only about 25 years ago. Hard to believe.
Technology is advancing more quickly than ever. That much is true. Voice recognition and AI are starting to cut into the jobs of transcriptionists, but only minimally at this point. I am just beginning to play with ChatGPT and other AI interfaces so I’m by no means an expert, but what I am experiencing though very impressive, doesn’t give me too much pause regarding the future need for live humans in transcription especially for more specialized transcripts for video production purposes shot with embedded time code–though optimization of that may be near on the horizon. Also, AI has a harder time with multiple speakers. While better than it used to be, differentiation between speakers is still a challenge for AI especially if there are more than two speakers.
And here’s the thing that I have kept emphasizing to people for nearly 30 years regarding speech recognition, and I stand by it now; dialect is a VERY complicated thing.
I grew up in Appalachia (yes, proud hillbilly here), and the dialect of Southern English I spoke in my hometown was definitely different than the dialect spoken by my cousins in the next town. This has been watered down somewhat by the internet age making the world smaller, but dialect can still differ, especially in more remote areas of the world, from town to town and in some cases community to community!
This is part of the reason you see ads everywhere about becoming a voice coach to teach AI. AI is still very much a work in progress when it comes to speech recognition. Where AI shines the most is word to voice translation. This is going to benefit the audiobook industry almost immediately as I believe authors/voiceover artists will no longer have to read whole books in cramped studios, although I don’t believe AI will completely take over the voiceover industry. I’m rooting for my friends who do voiceover. We still need warm “real” voices too.
I still assert that we will always need a real human to edit and QA any AI-generated transcript, especially for important documents like medical records and legal documents. There are some industries where perfect transcription is not as important, but it would be a shame for the standard of documentation to be acceptable of AI mistakes, bad grammar, and bad punctuation.
As I’ve always done I plan on embracing new technologies to enhance my own skills. If AI can help me produce transcripts faster and make me more money I’m on board. If one day AI takes my job so be it, but I’m not packing up my transcription pedal just yet.
The Scribe's Pen 