What Should You Do About AI Content on Medium?

Don’t waste time on solutions that don’t work. Use ones that do.

It’s natural for writers to worry about the impact of AI-written content. We worry about whether bots will replace us. We worry about how bots get the information they use.

However, this article isn’t about the legality or ethics of this technology; better writers have covered these issues on the site. This is about how we can deal with AI-generated stories on Medium.

Many of us are concerned about the volume of bot-generated content on the platform. We want to find ways to ensure that this content doesn’t detract from human writers and the work we do. We worry that our earnings will be affected by people who generate rather than write.

These are all natural reactions. We feel threatened.

However, our anger and anxieties don’t always help us here. We lose sight of what we can and should do. Sometimes, we waste energy on solutions that don’t work or that are actually harmful to our community. We then miss out on doing things that might make a difference.

Things you should know

Before we get into the nitty-gritty, there are some facts we need to know.

Medium cannot simply ban AI content

Medium isn’t necessarily unwilling to do this; it simply isn’t possible. There is no point talking about why they don’t just impose a ban.

They can’t ban AI. Nobody can.

How can anyone enforce a ban? While some copy/paste bot content is easy to spot, any writer with half a brain cell can take this content and make it look like it was written by a human. In skilled hands, it becomes indistinguishable from the content we write from scratch.

AI is here now. The stable door is open, and the horse has bolted at supersonic speed. We can’t simply collect it back up again and lock the door.

AI detectors just don’t work

Some writers have been calling for Medium to use AI detection systems. The fact is that there isn’t a detector out there that delivers 100% consistent results.

If there was, then everyone would be using it. We wouldn’t be having this conversation.

All detectors deliver false positives. So, innocent writers will be falsely accused of using AI when they don’t.

Don’t just take my word for it. OpenAI, the company that created ChatGPT, openly admits that they can’t build an accurate detection system.

They want to because that would make things easier for them on legal and government levels.

And, if you’re now rolling your eyes at me because you don’t believe they really want to do this, then think about all those ultra-skilled techy bods out there who have no affiliation to AI companies.

They could make gadloads of money by creating a detection system that actually worked. They can’t do it either.

Things you should think twice about doing

It’s natural to want to do something about our AI problem. If we don’t, then we feel helpless. However, there are some things that we shouldn’t do, either because they are pointless or because they are harmful.

Using detection tools

Some writers are using AI detection tools to try to weed out AI-generated content. As I’ve already mentioned, these tools don’t work accurately.

You might think that you can decide that someone is using AI if a tool tells you that they are or even just might be. But, you might be wrong, and you could do some harm.

How would you feel if someone ran one of your stories through one of these tools and you were flagged as an AI user? What might that do to your reputation and confidence? How could you convince your accuser that you aren’t using AI? After all, a tool said that you might be. Why should they take your word for it?

And, you should care about this. Don’t dismiss false positives as collateral damage. There is no humanity in collateral damage. We should be better than this.

Reporting AI users

We can report accounts if we feel they are breaking site rules. Some writers now report people who obviously use AI-generated content.

I’m not sure there is any point in doing this. In his October Partner Program Update, Buster Benson specifically said that we don’t need to report AI violations:

For now, you don’t need to report it as a violation of rules to our team.

I’m not sure that Medium removes AI-reported sites. It has other ways of dealing with them (more on this later). So, you’ll just get frustrated wasting time doing something that won’t necessarily deliver the results you want.

Reading/sharing AI content

If you read an AI story, then its writer earns money. If you share it with your friends, then they might read it too. They might even engage with the writer. The writer then earns even more money.

If you want to stop AI writers from earning money, then don’t read, engage with or share their stories. It really is that simple.

Things you can do

We’re not helpless here. There are some things we can do to deal with our AI problem.

Use site controls

While you might not think that Medium has an effective way to deal with AI-generated content, it is using its algorithm to do this. As far as I can tell, the measures it takes work in much the same way as Google’s search engine controls. Given that Google is the God of algorithmic development, this isn’t a bad thing.

In his update, Buster Benson tells us to use site controls to flag AI-generated content.

Our system’s ability to filter out AI-generated content is primarily based on content quality, so if you see content that you suspect is AI-generated, or any low quality content, be sure to take actions on it via blocking, muting, or “Show less like this.” That teaches the algorithm.

If you mute or block a writer, then the algorithm assumes that their content doesn’t have value. The same thing applies when you hit the ‘show less like this’ button.

The algorithm uses these prompts to help it set site visibility. It promotes good content and demotes bad content.

So, if you use these controls, an AI-generated piece will not be shown to as many people. It will sink below human-generated content. The writer will earn less.

Keep writing

I don’t know about you, but I enjoy writing less when I am stressed, frustrated and worked up. Up until a few weeks ago, I was getting really ticked off with AI writers. All that really did was make me less happy to write.

I’m not saying you shouldn’t be concerned about this problem. However, you should practice some self-care here. If worrying about AI is affecting your mental or physical health — or your ideation or writing — then try to let go a bit.

Fact is, if we write, then we keep populating the site with valuable human writing. This is what our readers want to read. We don’t allow AI writers to get the upper hand.

I’m no happier about AI-generated content than other writers. I simply think that it is better for me to focus on actions that can make a difference.

Obviously, your mileage may vary, and that’s fine, I’m not here to tell you what to do. However, if you aren’t sure about what works, then, hopefully, some of these points will help.