My computer has been invaded by AI.  

My laptop at this point is an extension of my brain. I cannot think unless I am writing or typing. Arranging words and thoughts. 

Yet now, every time I open an email, it’s to find an AI chatbot has already read my messages and summarized them for me. 

Yes, that’s right. Google’s Gemini reads my messages and repeats the basics in a summary. So now, if I want, I can read my messages twice. I can read the actual message, and I can read the summary message. 

I don’t see the point of this.

Likewise, whenever I open a document in an application like Adobe, a chatbot offers to read and summarize it for me. Mercifully I can click it away, but I do not know how to permanently delete it. What are these chatbot pests doing all over my computer? Where have they come from? Why are they offering to “read” and “summarize” emails and documents when I can do that myself? 

I did, wishfully, ask the Adobe chatbot to summarize a 20-page council document the other day. It summarized the first page or two. Everyone knows the good bits are buried on page 18. How is that useful?

Nonetheless, I worry that as a responsible professional I should be “integrating AI into my business operations.” I’m supposed to be “making it part of the team.” 

It is true that when you are running a newspaper, copyediting is a lot of work. Why is it a lot of work? Because no matter how many eyeballs you set to reading and proofing copy, mistakes slip through. 

The definition of “a lot of work” is when no matter how much you do, you are not done. 

Like the mountains of straw to be spun into gold overnight in fairy tales. Perhaps we need a Rumpelstiltskin chatbot. A magical creature who will come and do your mountain of work for you, asking only your first-born child in return. 

But wait. Isn’t that more or less what is happening? Except it’s not asking. 

So, I confess, for the proofreading, I went and tried it. I am a responsible professional after all. I found (in seconds of course) a proofreading AI extension called Grammarly. It took just a few more seconds to install it on my computer. Just one click and it was suddenly everywhere. It found its way to my desktop, where it asked permission to watch everything I do. It found its way into my operating system, where it asked if it could take over my computer. All of this in less than 2 seconds. It found its way into Word, the program I write in. 

When I opened a new document, there it was, blinking away, a little green G icon at the upper right. 

So, experimentally, I found and pasted a story we had already published — it is now the property of Grammarly; if you read the fine print, every single thing you do with a chatbot becomes its property — and clicked on the blinking icon. 

In one second it had found 29 errors in about 800 words. This is, remember, a piece proofread already by two different people and published. My heart started to pound. Well, to be honest, it was already pounding from the infiltration of my computer and the fine print. 

I started to go through the errors. Only to find they were not errors at all. They were what I would call style. Anything in the writing that was the slightest bit stylish had been underlined in red. The “corrections” suggested were all designed to stupefy the reader. Dumb it all down, take out any risks, render the sentences bland and repetitive. 

And wordy. 

The quotations, especially, the places people are free to sound like themselves, were redlined all the way. Grammarly does not like the spoken word.

I found that interesting. Perhaps because the chatbots have all been trained by “scraping” written — not to mention copyrighted — text available on the internet. They don’t have any idea how people talk. 

Anyway, you probably already knew this: a chatbot is not going to help us proofread, which is something we can do quickly, if slightly imperfectly, searching for mixed up letters and missing capitals and commas in the wrong places. Sure, we have the occasional worry about infelicities of expression, but mostly we are looking for actual typos. And checking dates and numbers and names one last time. 

The Word application already flags typos and even grammar in helpful ways. AI has been positioned for us as more sophisticated, of course, but this is not a sophistication that is useful. 

Like reading and summarizing emails I want to — and can — read for myself. 

Nonetheless, and this is the pernicious bit: keep an eye out for the number of times, and the sheer number of ways, you are told AI is something you ought to be using. 

The blinking. Green. Icon. Stare into it. Stop thinking. Hand over your emails. Blink blink. Stop reading for yourself. Blink blink. Stop thinking. Blink blink. Do it. Do it. 

The sheer number of “corrections” Grammarly offered on what was already good writing — lucid, concise, interesting — wastes far more time than it saves. I’m not the only one who has found this. There are reviews available on the internet, some by professional proofreaders. They are far more patient, detailed and comprehensive than my horrified mini-experiment, by people who tested and compared four or five AI offerings. All found the same thing. 

What’s really distressing to me is thinking about a student, in high school or university, or even younger. Someone in Grade Six. Or anyone not very confident about their writing — in other words, most people — running an essay or a blog post or a letter through one of these “proofreaders” and being told everything they are saying is somehow incorrect. 

What a crime. An attack on a fledgling mind.

AI is being foisted on us as a technology that can replace reading and writing for ourselves. Do not fall for it. And let’s not let our kids think they can’t write properly without ChatGPT. That it can read “more efficiently” than they can. All of this is not just nonsense; it’s a bizarre, self-consuming claim. Talk about eating your young.

Rumpelstiltskin only asked for one child. AI wants all of them.