Frequently Asked Questions about Spotting the Bot

Well, maybe not frequently asked. In fact, some of these have never been asked once, but I’m going to answer them anyway.

Table of Contents

What is this website?

This is a website that will teach you to spot content that was written by a large language model (LLM), aka an AI. It’s particularly focused on spotting writing from ChatGPT, the most popular LLM by far.

The articles here will walk you through the most common patterns that show up in AI-produced writing.

Who is this website for?

Great question. If you are a:

Teacher or professor looking to spot AI writing from your students
Student who doesn’t believe your teacher will realize you wrote your essay with ChatGPT
Publisher, marketer, etc., who thinks your audience can’t tell you’re outsourcing your writing to ChatGPT
Regular Joe who thinks you’re seeing a lot of AI slop, but aren’t sure / can’t express why you feel that way

… then I have a lot to say to you. Stick around!

Who are you and why should I believe you’re an expert in AI content?

I’m a professional writer and editor with about a decade of experience (you can see a brief overview of my work history on my profile). I’ve written or edited millions of words of web content, have hired dozens of freelance writers, and have reviewed application materials (resumés, portfolio pieces, etc.) from thousands.

A solid portion of my day job consists of reviewing text, such as freelance projects, and evaluating whether or not it was written by a human or an AI—something that’s gotten incredibly common since the release of ChatGPT. Of the freelance material that crosses my desk, I’d estimate a good 70%–80% is AI-produced at this point.

I’ve gotten good at spotting the patterns in ChatGPT’s output, and, just as importantly, explaining them to a skeptical audience. At work, that’s my boss; on this blog, that’s you.

In early 2025, I gave a guest lecture about this at National Yang Ming Chiao Tung University (NYCU) in Taiwan. It went fairly well, which gave me the idea to start this site.

Is it really possible to tell when a piece of writing was produced by ChatGPT?

Yes, it’s possible to tell the difference between AI writing and the real thing (without the use of automatic detection tools).

Unless ChatGPT is carefully guided (which it usually isn’t) or its output is painstakingly edited (ditto), the patterns in it are very strong and consistent. You can’t always tell with 100% confidence, but 90%+ confidence is often doable.

Caveat: This website is about spotting relatively clumsy, low-effort ChatGPT content, which constitutes the vast majority of the AI slop you’ll see online. Like I just alluded to, there might be much more skillfully produced AI content that gets past me, and I obviously can’t teach you to spot that, but it’s in the minority.

How long does it take to learn to spot AI writing?

I can teach you to do it in about 30 minutes. Start by reading this post on how to tell if something was written by ChatGPT.

Caveat: Getting really good at it—to the extent that you feel confident even about borderline cases—takes a lot of practice, but for 99% of people, “pretty good” is more than enough.

I read that spotting ChatGPT is all about looking for em dashes. Is that true?

If you’re already somewhat familiar with ChatGPT, you may have heard that em dashes (—) are one of its hallmarks. That’s sort of true, but only sort of; you’ll notice I’ve used em dashes several times on this page, and I didn’t use AI to write it.

There is a pattern to how ChatGPT uses em dashes (and other punctuation), but it is, unfortunately, subtler than “I saw a dash, ergo this is AI.”

Why are you so sure you’re right about ChatGPT’s patterns? Where’s the proof?

When I conclude that a piece of writing is AI, whether it’s a writing sample I’m getting paid to look at or just a post on an online message board, I’ll often do a little digging to confirm my suspicions.

Much of the time, I find some sort of smoking gun, such as a drastic shift in the author’s writing style right around when the first version of ChatGPT was released. I find this sort of proof often enough that I’m confident that I’m correct most of the other times, too.

Can’t I just use automatic AI detection tools?

Be careful about that. I’ve tested many AI detectors, such as Grammarly’s. I’ll say this for them: they’re getting better. They used to be unusable. Now, they’re … somewhat usable.

However, they still frequently return both false positives and negatives. I don’t trust them. I recommend that you be cautious about using them, too.

If you are an employer or teacher, please never fire somebody or give them a failing grade because your tool said their writing was AI. You need to use your own (human) judgment.

Also, realistically, if you’re not evaluating writing for your job (aka you’re a Regular Joe who just wants to be savvier about digital media), you aren’t going to plug everything you suspect is AI into a detector. You just aren’t. But you can develop the instincts to instantly spot ChatGPT content (and, if you’re like me, give it a rude gesture and navigate away).

Are you trying to sell me something?

No, I’m not selling an AI-spotting course or trying to get you to buy my AI detection app. There are also currently no ads or affiliate links on this site, and I don’t have a Patreon or tip jar.

Why are you doing this, if it isn’t for money?

AI writing is generally quite bad, and yet it’s getting more and more common.

I see ChatGPT’d posts on social media every day. Sometimes they get hundreds of likes or shares. I have seen major news outlets publish articles that were clearly AI-produced. I have seen well-respected literary magazines prohibit authors from submitting AI work (a good policy!) and then turn around and use AI on their own websites, confident that no one would catch them.

I find all of this frustrating. I believe AI is a valuable tool, but I do not agree with using it to fill up the Internet with forgettable garbage, which is how it’s being used now.

I can’t stop this trend, but hopefully, I can teach a few people to not be taken in by it.