Medium Asked A Question I Don't Think We Can (Or Should) Ignore
I don't need to read 400+ comments to know what I think. Plus, a thank you and a writer I'd like you to meet. :)
Happy Friday,
In September, Tony posted on the official Medium blog to say Medium is defaulting to “no” to AI training on our stories. That’s their official stance. For now. Which means for now, AI cannot train on writing posted on Medium.
At the bottom of the post, he asked two questions…
Yesterday, I got the Medium October Newsletter. There they were again, the same two questions. With a link back to Tony’s post.
One writer said her concern is this — how would they enforce what is scraped and what isn’t? That would be her concern. That’s why she would leave.
Here is Tony’s reply…
“FWIW, I think enforcing what gets scraped and doesn't is easy and straightforward. That has known solutions that would be easy to incorporate into any deal. Then it's just up to us to provide you a setting and up to you to set that setting.” (source)
I don’t buy it.
Sorry.
I don’t think it’s easy or straightforward. If it was, there would be no scrapers or spammers or hackers on Medium. But there are.
Let me show you something. Open this link in a new tab.
Look at the last two lines.
Here’s a picture of it.
That’s a robots.txt file. A robots.txt file tells robots what they can and cannot access on a website. The last two lines tells GPTBot that it’s disallowed. It can’t access anything on the site. Any other bot? Yup. That’s okay. It can go right ahead.
So, what if the AI company decides to come up with another bot with a different name? Well, then Medium has to disallow that one, too. And if they come up with a third and fourth bot and fifth bot? Well, they have to “disallow” those, too.
That’s not wackadoo theory. It’s how bots work.
But, what if Medium allows “some” content to be scraped. Well, then they need those permissions set for every writer account on Medium.
And if the AI bot keeps creating new bots? Well, then they need to set the permissions for every bot, for every account. Yes for you, no for me. For every bot. For every user.
And you want me to believe that’s easy and straightforward?
Sorry. I don’t. I have been the one working past midnight to keep bots from scraping websites for my clients. Chasing down bot names and editing that damn file and then checking user logs to see if it’s stilling gobbling, gobbling.
I have worked in tech for twenty years.
Spend ten minutes googling phrases like “what is a scraper bot,” “scraper bot” and “website content scraper” and you cannot possibly come away from that exercise thinking it is “easy and straightforward” to manage what a scraper bot does.
I have a different question. Why would Medium entertain the idea of letting AI train on stories they don’t even own the rights to?
Especially when most of us say no.
Because we do say no.
Loud and clear. Here’s another quote from Tony’s post…
When we surveyed our authors, 92.2% of you said that you want us to take active measures against AI companies until these issues of fairness can be sorted out.
The “issues of fairness” he refers to are the 3 Cs. Consent, credit, and compensation. When over 90% of writers surveyed want AI to keep their hands off, why is this even a conversation? That’s what I want to know.
That’s before we address compensation. You want me to think giving us 10% more of what we earn now constitutes fair compensation? So an author who earns $200 gets and extra twenty bucks to turn their writing into an all you can eat AI buffet?
And how much more do the top writers get? How is a percentage of our earnings fair compensation for feeding us to the corporate AI machine?
There are lots of places where it’s great to be on the cutting edge. Letting AI robots get their hands on our writing for peanuts is maybe not one of them.
I don’t think we can ignore this. Here’s the post. If you haven’t read it, do. And leave a comment. Add your voice. Your opinion matters.
A thank you and someone I’d like you to meet…
On Reflection is now my fastest growing publication and gets more views than History of Women, my oldest publication. Over half our posts at On Reflection get boosted and much of that is thanks to my co-editor, who I’d like you to meet. Couple of good reads I think you’ll enjoy…
Through the Lens — a delightfully goosebumps kind of story
Fox’s Garden — a visually stunning book review worth reading
Poke around. I think you’ll enjoy.
If you enjoy my writing, please click the heart or share this post. Thanks. :)
xo,
Linda
I'll tell you what...I'm just about done with Medium. I don't trust the AI situation, and my earnings are down 80% this month. I don't trust them as far as I can throw them. Why should I keep giving them content?
That's a powerful "heads up" Linda, thanks. So if this A.I. invasion is inevitable, as it seems why can't we as writers "throw a wrench" in the pudding as a manner of speaking? Pepper your writing with a selected, or a few selected irrelevant words that to humans make no sense at all when you're reading the work, but A.I. will not know that those "word mines" are there to thwart their leaning, they will include our stupid words and spit them out in A.I. produced work hahaha. Someone could even write a program designed to insert these words into your writing with the click of a button? And you could put a note of warning above your work informing the reader of A.I. buster words, maybe under your subtitle in parentheses? I don't think it's a good idea for fiction, but non-fiction it may work for and I'm not sure how most readers will react, maybe think it's even funny reading, or maybe just irritates them. Hey I"m just brainstorming, don't pay any attention if this is too stupid to even think about, but it seems to me writers have to take matters into their own hands rather than just hoping platforms will protect them.