It was intense, not the least because we decided to completely redesign our processes and migrate to a new software package. Probably not the best idea.
I also used my friend co-pilot. Microsoft’s AI tool.
It’s the first time I’ve used it significantly to help with writing, and it was interesting.
To demonstrate, here’s how AI summed up the experience: "In the realm of quality management, the journey towards excellence is a continuous process, punctuated by significant milestones. One such milestone was achieved last week when we successfully underwent our triennial recertification audit for the ISO 9001 quality management system. This event not only reaffirmed our commitment to maintaining the highest standards of quality but also provided us with an opportunity to reflect on our accomplishments, learn from our experiences and plan for future improvements. This article aims to provide an in-depth account of our audit experience, shedding light on the rigorous processes, diligent preparations and the invaluable insights gained, which collectively contribute to our ongoing pursuit of quality excellence."
Have I lost you? Not very summed up is it?
And the fluffy words — pursuit, triennial, punctuated. I wish they’d dropped an exemplar in there too.
I gave it another go, this time adjusting my prompt to ask co-pilot to write in my own style by referencing columns from the ODT: "In the high-stakes game of quality management, we’ve just rolled the dice and come up trumps! Last week marked our ISO 9001 recertification audit, a three-yearly event that’s as nerve-racking as it is necessary. But this wasn’t your run-of-the-mill audit. Oh no, we’d decided to spice things up a bit — new software, revamped QMS, the works! So buckle up, dear reader, as we delve into the nitty-gritty of our audacious audit adventure. It’s a tale of tenacity, teamwork and top-notch quality management, Otago style!"
Obviously, AI thinks I write in corny cliches and buzz words. Sorry "dear readers" — that bit made me sound a bit like Lady Whistledown.
What I learned from the experience was this: AI has a great place in summarising content and data.
In many cases I "brain dumped" out our content and asked AI to summarise it into matrix tables or bullet points grouped to address each requirement of the applicable standard.
I also had a crack at a flow chart AI tool, but it was a squiggly mess — maybe it was trying to tell me something.
AI is an excellent research and educational tool — if you know what you’re looking for. I worked through the ISO 9001 standards clauses and prompted AI to review our content for comprehensiveness and best practice.
It would review and suggest areas that were missing, along with links to sources for further research.
While I did use it for several statements, I found that unless I’d done 80% of the writing already, it really was an expert at saying a lot without saying much — a writing style I find particularly frustrating.
It’s all about the prompts, especially tone of voice. If you want an authentic and engaging voice with your audience, particularly if that’s your team, I personally don’t believe AI can do a better job.
You can use AI, but you need to put the time into training it and you must establish your own voice first. Be especially careful asking it to write in a "playful" or "down to earth" tone, as it can be downright corny: "Hey team, guess what? We aced our ISO 9001 audit last week! It was like a roller-coaster ride with new software and a shiny QMS. But we did it, all thanks to our super teamwork. So, high five everyone, we’re quality champs, Otago style!"
I do also worry about plagiarism, as in many cases when reviewing the source material it was obvious that AI, while avoiding direct verbatim copies, it came pretty close.
I’m also really careful about what data I put into AI and what tools I use: I would never put confidential or company-sensitive information in and tend to get it to write anonymously.
But perhaps my greatest concern is that there is no checking by AI on the validity of the sources it’s referencing — how do we know the source data isn’t erroneous?
With this lack of discernment, how long will it take for a critical mass of content to have been written by AI and AI starts to source itself? This is why it’s so important to use a tool that references sources.
I asked co-pilot about this point and here’s what it said: "It’s important to note that while AI can produce text quickly and consistently, human oversight is essential to review and refine AI-generated materials. AI-generated content may lack the human ability to verify facts accurately. While AI algorithms can produce text by compiling information from multiple sites and based on learned patterns, they may not always distinguish between accurate and erroneous information."
My parting word is this. I’m no AI expert: I’m sure all of the above has been well documented and debated by the experts and so I simply bring a laywoman’s opinion and observations. Hopefully, it’s prompted others to do the same thing.
- Sarah Ramsay is chief executive of United Machinists.
