Does Turnitin Detect AI Writing Tools Like ChatGPT & QuillBot?
6 mins read

Does Turnitin Detect AI Writing Tools Like ChatGPT & QuillBot?

The explosion of AI writing tools like ChatGPT, Google Gemini, and Claude AI has been a godsend for slackers across the globe. Now, for the first time in human history, we can get the machine to do our homework for us. God, we did it. A proper utopia.

Not so fast, kid.

As these AI writing tools evolve, so too do the tools that detect AI writing. Which is bad news for any potential cheaters out there (don’t worry, we know you wouldn’t do that). Essentially, it’s like a game of cat and mouse, where both the cat and mouse are advancing at exponential rates, producing results that would have literally seemed like magic a decade ago. Oh, and one is far better than the other. But we’ll get to that later.

So, can Turnitin truly sniff out your AI-assisted masterpiece, or will you (hypothetically) get away with it? Buckle up, Bucko, we’re about to find out.

What is Turnitin?

If you don’t know what Turnitin is then I envy you, truly. It means The Almighty has blessed you with a good life. Basically, it’s software that checks your essays, assignments, and whatever else you feel like feeding into it for plagiarism. It’s there to make sure you haven’t been copy-pasting your way to an A. 

It compares your work to a massive, ever-growing database of online content, academic papers, and previously submitted student work. Basically, if you’ve been a bit naughty, Turnitin will know. 

Well, that is until AI came into the picture.

The Rise of Bot

Look, for a minute there, the best ChatGPT could do was write you a fun poem about being in love with Shrek in the style of Donald Trump. And it was bad, but funny, and new! So we all had a laugh playing around with it, and some (idiots) used it in professional settings, which was like asking a 12-year-old to do your taxes. Sure, they could technically do it, but it would be bad.

Then, we blinked, and the machines became harder better faster stronger. Now, the temptation to use particularly eloquent models like Claude and ChatGPT 4 and Gemini Advanced is tempting because, well, they aren’t too bad now. With the right prompts, they spit out some decent shit — shit that’s potentially better than what you could produce (if you suck).

So, can Turnitin catch you using AI?

Well, that depends on who you ask. According to Turnitin, its software has been “independently shown to have high effectiveness in correctly identifying AI-generated content, when compared to other commercially available detectors”.

And they have the numbers to show that alleged AI content is being flagged. Between April 2023 and October 2023, Turnitin claims more than 130 million papers have been processed for AI detection. Of those papers, 3.5 million were flagged as containing 80% or more AI generated content. When you lower the bar to 20% AI content, that raises the number of papers to 12.5 million.

But is it reliable?

Well, honestly, no. Turnitin says its AI writing detection works by analysing patterns that indicate the use of AI writing tools, but it doesn’t actually say how it goes about doing this. Some universities, like Vanderbilt University in the US, have decided to disable Turnitin’s AI detection tool for the foreseeable future, citing its unreliability.

When Turnitin launched its AI detection feature, it claimed it had a 1% false positive rate. Think of the tens of thousands of academic papers any one university processes every year — a 1% error rate is wild. That’s potentially hundreds of students a year going through the life-shattering process of being wrongly accused of cheating. Now, scale that to every university on the planet that uses Turnitin (that’s more than 16,000 academic institutions, mind you). 

That’s a hell of a lot of potential students’ lives being completely turned upside down by false allegations of cheating by clearly flawed software.

AI detectors are racist (no, really)

AI detectors aren’t just bad at their jobs; they’re also biased against non-native English writers. According to a study published last year, AI detection tools consistently misclassify non-native English writing samples as AI-generated. This is obviously bad! Bad robot!

What happens if you get pinged?

Look, I’m not your mother, and I’m not here to tell you what to do with your sad life. If you choose to get a little help from the machines, just know there is a very real more than likely chance you will be picked up on it. Because, even though it is clear that AI detectors aren’t actually reliable, universities and the dinosaur professors that roam their grounds grumbling about this lazy generation while struggling to attach PDFs to emails still insist on making use of them. They don’t seem to mind that AI detectors flag the fucking constitution as being written by AI.

Though imagine the simulation-breaking plot twist if it actually was lol. God, that would be so sick.

Anyway, the long and short of it is that Turnitin can probably detect if you’ve been using AI in your essay—but it can also ‘detect’ AI that straight up isn’t there

You could cheat your way to the top, and the machine mightn’t blink an eye, or you could do the hard yards, study your little heart out, and still be called a dirty little liar by a robot. Sucks, I know. But hey, apparently, that’s good enough for Turnitin. 

Good luck out there.