The Dead Essay Theory: How Chat GPT is Killing Academic Writing

 I have seen the term “dead internet theory” floated around since the debut of Chat GPT. The theory goes something like this:

• Stage I of the Internet: only the select few could create content, it was all human generated, and mostly accessed with web browsers.
• Stage II of the Internet: anyone with an internet connection can now create content & put it on the web (websites, blogs, videos, pictures, social media, etc.) and increasingly large amounts of traffic are from mobile devices and apps, not just desktop based web browsers.
• Stage III of the Internet: generative AI enters the chat, much of the web’s traffic is actually bots, and LLM’s not only create content (AI Influencers, anyone???) but also comment, respond, & interact with other bots.

According to some, this means the end of the internet as we know it. Is it inevitable? Is it already happening?

I’m not in the habit of making predictions (us history teachers know better!), but I’ve been thinking all summer about teaching writing in school. I don’t claim to have any answers, but many of us humanities and language teachers have started to see the problem:

Assigning Essays in the Generative AI Era

If you need me to explain how this is a challenge, you must not be an educator, writer, or work in schools. But for the rest of us, the arms race between tools like Chat GPT & AI detectors like GPT Zero are exhausting, and ultimately, futile!

If you are under the impression that AI detection tools are the answer to students having access to generative AI tools in Snapchat, built into Google Docs, Chrome extensions, and everywhere else… you are going to be disappointed!

How do I know? I’ve tried using them. My conclusions?

1. They are labor intensive
2. They are unreliable
3. You cannot base a decision to fail or accuse a student of academic dishonesty with their results

You can look up the independent testing results on all of these tools, but I’ll give you the shortcut: most people never heard that Open AI, the company behind Chat GPT, had their own AI detection tool. HAD.  They don’t HAVE it anymore. Why not?  Here is what it says on their now defunct product page:

So you’re telling me the company that is building one of the most powerful LLM, and certainly the most widely used one, can’t even build an AI detection tool?

Yep, and it wasn’t even close. 

So what makes you think GPT Zero, Turn It In .com, and all these other ed-tech companies can do it better than the creators of Chat GPT?

Let's be pragmatic: even if they could accurately detect AI, do you want to play this cat-and-mouse game between technologies & trying to bust students for “cheating”?

I don’t. I tried it. It doesn’t work.

Enter the Dead Essay Theory:

I don’t want student writing to turn into AI generated content, which I in turn use AI powered tools to police their AI usage, or worse, AI tools to give feedback & comments to their AI Bot from Snapchat or Meta…

OK, so if banning AI or using detectors isn’t the answer…what is?

Honestly, I don’t know. But I know there has to be a better way. I’m excited to keep experimenting (responsibly!) & iterating new ideas with my students. 

Despite the onslaught of “Never AI!” teaches commenting on my social media posts, I’m going to keep finding new & better ways for my students to learn, including with AI tools.

If you have the same uneasy feeling about assigning “Dead Essays” to your students, I encourage you to connect with me! 

If you are a student stuck in a class with professors who try to "ban" AI or tell you Chat GPT is cheating, don't listen to them! There are tons of ways to use AI tools that even the most strict teacher couldn't accuse you of cheating with.

Leave a comment (even if you disagree!), DM me, or follow my social media content. My email is [email protected]