A notion on AI usage
A personal addition to the ongoing debate
I’ve spent an awful lot of time reading people’s notes, essays, and articles on opinions for and against, and everything in between, over the last few days. Fascinating stuff.
At first, I was indifferent to this new AI scanning tool. I thought it was great for those who wanted to know whether an article (and I’m thinking non-fiction here) was written by AI, or to what extent. I still think it is useful in many cases.
I also think transparency is a good thing. Those who use AI openly can explain their workflow. Well-crafted essays and articles by a human using AI tools to produce their work are obviously worth reading — it is still their work.
Then I read about the concerns of many writers, mainly fiction authors, who had personal experience of Pangram flagging their human-written work as AI. This is where my indifference shifted.
Being falsely labelled is not ok. It doesn’t matter that you can contest the result. We all know how this works.
Being accused and convicted in the same sentence simply for turning off the option to scan your work is also not ok.
Will this tool actually achieve what it was meant to do — reduce AI slop that no one wants to read? I’m not so sure. Could we not simply choose not to read what we deem crap, deciding what “crap” is with our human brains?
Open communication, transparent negotiation and consent matter in all areas of life. Respecting these principles is hugely important in any type of relationship.
The way Substack quietly changed their privacy policy in May, then announced and introduced a tool with no prior notice or option to pre-consent or opt out, was neither transparent, open nor respectful.
This bothers me immensely. The false positives bother me. The tool’s proven biases bother me.
I am not anti-AI. My husband is in IT, is very pro-AI, and makes great use of it. I don’t, as I feel the craft and the manual process bring me joy, and automating and optimising would take that away. Still, I think AI is good for many purposes and in many fields.
I think there is AI slop as well as human slop, and there is great human writing and great AI-assisted writing.
In the case of fiction, and since there are false positives, I’d urge everyone to listen to their inner voice rather than to a percentage in a pop-up window. If you felt like reading an author’s works was worth your time before this function came along, I don’t think Pangram should change that feeling.
THAT would be outsourcing human intelligence to AI.
If you’re interested — here’s how I work
I use Gemini for much of my research because it saves time. If I use facts she presents, I check the sources when relevant to assess their credibility.
I also use Grammarly to spell-check my writing. I use almost no commas when writing my first draft, and I’m grateful for Grammarly to add them where they belong.
I’m a non-native English speaker, reader, and writer. Although my entire 15-year professional career was in English, it still shows that I speak 4-5 other languages in my day-to-day life. I could write a novel about how, but I won’t do it here. This is why I accept Grammarly’s suggestions to improve sentence structure and flow.
I also sometimes pick a synonym that Grammarly suggests because it saves me time. I like to rewrite countless times; I edit and polish endlessly, and I would have changed the word in one version or another anyway. I might even change Grammarly’s suggestion along the way.
I honestly thought that using Grammarly this way was not_using_AI. I told my husband a few days ago, and he informed me that I’m dead wrong. LOL.
So I had to change my declaration from “not using any AI” to “not using generative AI”. I admit it is more accurate, and the only reason I added this note to every profile description in the first place was my love for and overuse of em-dashes, which, some time ago, led to a sentence without a trial. That storm has now passed, it seems.
I write my prose and essays, my articles, blog posts and scenes from scratch in a Word document, then I copy-paste them into the platform where I publish. My words are my own, and using Grammarly and Gemini as I do does not change that. I mark citations clearly and always give credit to the author I quote.
I’ve chosen not to allow you to scan my work because, at the time of writing, I strongly dislike this new tool. I don’t believe it should be incriminating to exercise your/my right to NOT CONSENT.
I also agree with many who have expressed outrage at “How I make this” — no author or artist needs to reveal their creative process if they don’t want to.
I have outlined how I work here, and I have done so willingly. If you do not trust my word, if you have suspicions and feel you want a percentage to evaluate whether you should read my work, I think we’re a poor match, and I kindly invite you to read elsewhere.
(I’m, of course, dying to know how much of a Robota I am, so I scanned this one, and it came back 100% human.)
I will admit one thing concerning AI, that it has uses as a quick reference or a lore tracker. In my current series of stories, I have many pages of notes, languages, cultures, fake histories, character names, etc. I have fed it all into one of the AI programs. It is a helpful review tool. I can feed my chapters into it, to review to make sure I haven't accidently used the wrong character nickname or location. Sometimes it doesn't catch mistakes, and they are things that only careful editing really catches. But for simple 'does this chapter fit with what I previously wrote about XXX', it can be a useful tool. The stories are still mine, the writing is still mine, and even the references are mine, but the program simply helps me manage them. That's all.
yup 💯