23rd July, 2025
If you’re reading this on a handheld device, content will display best with the device held horizontally.
Is AI putting words in your mouth and deliverables?
THIS ARTICLE includes reporting on my past surveys about AI in the assistant career, and a new multiple choice survey that should take no more than a minute or two to complete. Please add your voice and share the link to this page, as we continue tracking the extent to which assistants around the globe are using GenAI, and how you're capitalizing on it.

we’ve come a long way
Next Wednesday marks 32 months since the public launch of ChatGPT. We’ve come a very long way since December 2022, when I first wrote about ChatGPT. Not only have we seen the emergence of additional AI/GenAI resources; the extent to which people turn to AI has snowballed.
In January 2023, when I surveyed readers about whether you’d begun exploring ChatGPT, respondents were evenly divided in three camps. One third of respondents had already established ChatGPT accounts, another third had been reading about it, and the remaining third had not yet explored what was a very new tool.
I’ve been writing about digital innovation and disruption, both of which ChatGPT represented, for years now. Knowing from firsthand experience that people have varying appetites for change, I asked how you feel about new technology and change in the workplace. I posed the question to three survey groups, and you’ll see the results below. In which camp are you?

DO YOU CHOOSE TRADITIONAL SEARCH ENGINES, OR AI?
I also asked readers in January 2023 if you could see yourself turning to ChatGPT rather than Google or other search engines. These were very early days, with Bard (now Gemini) and other options not yet publicly accessible. ChatGPT’s knowledge base at that point in time extended only as far as September 2021. It made sense, then, that 67% of respondents considered it unlikely, while almost a third said you might turn to ChatGPT over Google if you could trust its accuracy.
In 2025, many routinely turn to ChatGPT, Claude, Copilot, Gemini, Grok, Perplexity and other AI resources with questions or searches for which you may previously have turned to Google or another search engine. In 2025, even when you do choose a traditional search engine for your query or search, it’s highly likely the response topping your screen will be an “AI Overview”.
Now, AI overviews commonly top search results; we’re being coached
Shelagh Donnelly
We are, in effect, being coached or trained to rely on AI even if our intent is otherwise – and, in many instances, it’s working.
On July 17/25, I published a couple of LinkedIn posts relating to the use of AI to create rather than edit posts for that media platform.
Among the comments on my public post (I also publish posts to Lucy Brazier’s Executive Support Magazine LinkedIn group) was rather disheartening word from someone I respect; she has an impressive track record and I’ve not known her to waste peoples’ time. She wrote, “I don’t enjoy hearing from others that I should stop asking them and other humans questions but rather ask AI instead.”
AS AI EVOLVES, more ARE TRUSTING IT TO COMPOSE, EDIT
I know this anecdotally, from conversations with assistants. As well, I posed questions about your GenAI usage in February 2025. At that point in time, over a third of respondents reported using AI daily or almost daily to draft workplace communications or deliverables. Almost half the respondents reported doing so occasionally. As you’ll see below, some employers restrict GenAI usage to Copilot.

It’s unsurprising that 10% of respondents in February 2025 were not using GenAI for communications and deliverables. As mentioned in earlier articles, there are workplaces where the nature of an organisation’s operations is such that risk management practices preclude using GenAI. There are also workplaces in which boards and management teams have taken – or continue to take – a cautious approach before dipping toes too far into the AI waters. Elsewhere, budgets and strategic priorities can be a factor.
“5 Types Of LinkedIn Content To Avoid Posting In 2025”

That’s the headline from a Forbes article this month.
The writer told us LinkedIn’s algorithms have caught on to an emerging practice some have adopted: the use of AI for content creation. People turn to AI to write posts for publication on LinkedIn.
The article encouraged readers to draw on AI to help with structure, while ensuring the substance reflects the human’s expertise.
There were stats and links to additional stats about reliance on AI in writing for LinkedIn, yet the article pointed out what more than a few actual humans picked up on ages ago: a number of LinkedIn posts and articles reflect similar phrases, formatting patterns, calls to action and even click bait themes.
Sometimes, as Mallory Rothstein, one of my counterparts, pointed out in conversation months ago, posts seem to – as a result of reliance on AI – unwittingly regurgitate other peoples’ published work.
Authenticity is important to me; you’ll see I’ve highlighted it near the top of my LinkedIn profile. I do use GenAI and, while I rely on my own noggin, experiences, research and observations rather than AI to write what you see from me on my website and elsewhere, I get that AI can be a practical tool to support writing. All this leads us to the quick survey you’ll see below.
WHERE DOES GENAI FIT INTO YOUR CAREER AT THIS POINT?
Here’s our mid-2025 check in on how and if you’re using tools such as ChatGPT, Claude, Copilot, Gemini, Perplexity or other resources to help with writing or editing communications.

Leave a Reply