Copyediting and AI: Are we there yet?
Many publishers are interested in pursuing a copyediting workflow that incorporates artificial intelligence (AI). The biggest questions in 2023 involve what works and what doesn’t.
While tremendous strides have been made in Natural Language Processing (NLP) that enable machines to analyze and process language data, this is quite different from actual understanding. Understanding or comprehending language will require achieving singularity (also known as Artificial General Intelligence). Estimates vary regarding when this will be reached, but most experts agree that we are decades away.
The majority of AI CE tools currently rely on various forms of pattern matching algorithms, which look for exact matches with pre-existing patterns. At their current level of sophistication, such tools can be useful in many contexts and may indeed work a fairly high percentage of the time.
But there are aspects of language that require advanced cognitive abilities: The tool needs to not only recognize a pattern, but to comprehend the context in which a pattern appears or any given word is being used.
An example of the difficulties such tools face in copyediting would be article usage: when to use “a,” “an,” or “the.” Because what is correct is based entirely on the context and since the context can change even within a sentence, the tool must comprehend how it’s being used. Most AI experts agree that the technology is many years from reaching that level of sophistication.
Book a meeting with our resident copyediting experts to determine how a hybrid copyediting model can work for you.