While generative AI is highly capable, it has limitations and may sometimes produce incorrect or nonsensical responses. Therefore, it is important to critically evaluate and verify the information provided by any AI system. This was a key point raised at the latest Coffee Club of Safrea’s Eastern Cape region.
The small group watched the recording of one of PEG’s 30th Anniversary Conference presentations called “Generative AI and its impact on editors,” presented by Dawn Green and Richard Steele. The presenters provided an overview of generative AI and tested the leading AI chatbots (ChatGPT, Bing and Bard) with editing and proofreading tasks. These included punctuation, grammar, plain language, neutral and sensitive language, generating macros, changing English conventions, changing reference conventions, and proofreading. The outcome produced very useful, but also less useful results.
Negatives and positives
The negatives of generative AI are known and written about such as plagiarism, cheating, inaccuracies, bias, and harmful use – especially since the industry is unregulated. The positives offer great potential, especially from a research perspective. AI also provides inspiration and explanation, it executes instructions and gives people free access to vast information, allowing them to improve their critical thinking ability as they have to assess the results. Time-saving is the big positive as AI can help us to answer emails, write contracts, rewrite sentences to make sense, and do much more. But, ultimately, we are in charge!
Functional tasks and your own voice
A guest presenter, Kevin Walker, joined the presentation as a full-time freelance editor. He tested non-editing functional tasks among the three chatbots such as business admin, correspondence and the writing of “difficult” or repetitive correspondence. He compared their output in terms of style, tone, length, tact, politeness and competence or accuracy.
Kevin came to the conclusion that all three chatbots provided different angles you could choose from to form a response. “The instructions certainly matter – you must be precise: what you put in, you get out. AI reduces mental work, but you have to carefully check for the truth, clarity, and especially your own voice to come through.”
Kevin’s recommendation is to experiment and have fun with the chatbots. “The more we play, the more we’ll see their limitations and get the reassurance that we will not be replaced by machines any time soon.”
Developmental editing and human intervention
Another guest presenter, Dr Linda Pretorius, tested the chatbots from a developmental editing point of view. She defined developmental editing as using and adapting the existing parts of a manuscript to shape a cohesive, coherent text that works like the author originally intended the reader to understand it. She also compared it to the definitions of three other developmental editing experts.
After running a few trials with the chatbots, she came to the conclusion that AI is more of a hindrance than a help. This is due to the common element in all four of the developmental editing definitions – it requires human, creative, and critical evaluation, which generative AI cannot provide.
How do you stay relevant in the face of generative AI?
Know your competition, play around with generative AI, and keep doing so as it is constantly changing and improving. Assess how it could potentially help you in your editing business and find innovative ways to increase your knowledge and expertise to stay ahead.
Abbreviations:
• AI – Artificial Intelligence
• PEG – Professional Editors’ Guild