Will AI Solutions Replace Stenographers?

By: Verbit Editorial



Popular posts

Adding Captions To Instagram Reels & Videos Adding Captions To Instagram Reels & Videos
Adding Subtitles in DaVinci Resolve Adding Subtitles in DaVinci Resolve

Related posts

How do executives feel about AI? How do executives feel about AI?
LRT logo
Verbit launches revolutionary real-time transcription service for legal proceedings  Verbit launches revolutionary real-time transcription service for legal proceedings 

Let’s start with a basic, but important question: What is AI? Essentially, it’s the broad discipline of creating smart machines that can perform tasks that are reminiscent of human capabilities such as recognizing objects, texts, sounds, speech, and solving different types of problems. Another more technical term for AI is machine learning. 

AI is often mentioned as being superhuman in terms of its capabilities, so a quick reality check on what AI can and can’t do wouldn’t hurt. Two decades ago, IBM designed a program called Deep Blue, a software that was able to defeat world champion Garry Kasparov in a chess match, a feat that shattered the way people thought about machine capabilities. Fast forward to 2018, where another huge success in the domain took place. AlphaGo, designed by the DeepMind group in Google was able to defeat the world champion in “Go”, a much less rigid game than chess. It was predicted that it would take many more years for a machine to beat a human in this type of situation, but it happened. 

But which situations result in AI underperforming? What is different about those tasks, compared to a task where the AI excels? The difference is real-world knowledge. For example, accurately identifying an image requires an understanding of how things exist in reality, such as how objects look when they’re rotated in three dimensions. It’s a totally different set of skills than logic-based problems like chess or Go. For situations that involve real-world knowledge, humans maintain a huge advantage over machines. 

Therefore, in the world of court reporting, we’re faced with a dilemma: How can we still leverage machine capabilities in a domain that is obviously dependent on knowledge of the real world? The answer lies in the hybrid model, combining the strengths of both artificial and human intelligence for optimal results. This involves deploying both machines and people, each in their own element, to carry out tasks in the most efficient way. This is done by assigning the AI the easier, more repetitive tasks, which it can perform accurately, quickly, and for a low cost, and leave the more difficult, complex and creative tasks for highly skilled individuals to work on.

Let’s dive into the example of legal transcription in greater detail, and how both humans and machines are involved in the process. A computer-generated transcript can be produced using speech-to-text technology. However, in the legal domain, there are often complex and industry-specific terms mentioned, meaning the computer may mistranscribe some of these terms. That’s where the element of human expertise comes in. Human transcribers can go over the automatically-generated text and make any necessary corrections. Legal transcription is a perfect example of the fusion of artificial and human intelligence, as the computer does most of the simple work, with human expertise coming in later to perform tougher terminology and context-related corrections. 

Even in spite of complex subject matter, there are ways to ensure that automatically produced transcripts are of high quality. The last decade has seen tremendous advances in speech recognition due to huge amounts of data becoming available. Developments in machine learning, deep learning, and neural networks have enabled significant improvement, although a gap of understanding still remains. Then there is the issue of audio elements. If there is difficult audio, it’s challenging to obtain accurate transcription. Similarly, if the speakers have accents that the machine wasn’t previously introduced to it will not perform at its best. 

The best way to minimize these issues and ensure the highest level of precision is to train the machine on as many elements as possible, most notably on taxonomy. This is where AI really shines. In the domain of court reporting, if the topic of the hearing is known in advance, such as a medical issue or an insurance claim, then this data can be fed to the AI. This will bring in the necessary terminology, have a specific model for the case and, consequently, produce much higher accuracy for these terms than a human, who likely will not be particularly familiar with these concepts.

Beyond the process of transcription, AI can be practically applied in a variety of tasks related to court reporting. Think of a live court reporting session. Suppose there is a need to go back to what someone said earlier in the proceeding. A stenographer would have to search their notes for his statement, while a digital recorder would have to quickly scan the log notes, hoping they wrote something relevant. On the other hand, an AI program could simply search for the phrase that is entered and find the exact place in the audio where it was stated. Even if the original transcription contained an error, the technology would allow searching the audio itself, which could then be played back. 

Let’s look at another scenario. It’s the digital reporter’s duty to make an accurate record of the court proceeding, so if someone speaks unclearly, they must mention it. This is something that can sometimes go unnoticed by a busy human but is easily detected by a machine that can pick up on it and alert the recorder. The same goes for going off the record, ambient noise, and overlapping conversation. All of these scenarios can be detected by machines which could then alert the reporter.

Therefore, going back to the original titular question: Will AI replace stenographers? The short answer is no, as human expertise is essential to work together with technology. However, in order to maximize the use of their unique and specialized skills, stenographers should target situations where technology cannot be applied, such as scenarios where there is no digital setup. 

The rise of AI technology has significantly impacted almost every aspect of daily life and nearly every professional industry. AI offers distinct advantages and strengths in many domains, and legal is no exception. In particular, AI and machine learning technology have the potential to achieve faster transcription turnaround and a high level of accuracy for court reporters, as well as assist with other elements of a court proceeding.