Here’s a place where you can see my notes about AI writing. I’ll be updating this in the coming months and eventually hope to include pedagogical takeaways.
If you’ve never taken a look at what AI writing looks like, I’ve put together a few samples. To make each of the samples on this page, I gave the AI a prompt about using AI to write college papers. Each sample shows the result of asking it to write in a different genre (blog, essay, social media post, press release, creative stories). I’ve added some annotations to each with my thoughts on the writing, mostly focusing on what tendencies I notice. Here are a few caveats:
- This isn’t a systematic review of how AI writing bots work. Everything here is anecdotal.
- These posts were written both using Notion AI and ChatGPT. I think ChatGPT is probably a bit more sophisticated. You’ll notice that each of the samples shares a very similar structure, regardless of the genre I asked it to write in. You’ll also notice that a lot of the sentence-level language is similar sample to sample. From my experience, ChatGPT is a bit less deterministic than Notion AI in how it writes essays (i.e. each time you query the ChatGPT AI with the same prompt, you get a unique response) but you can find recycled patterns in both.
- The most immediate difference between Notion AI and ChatGPT (at least to me, a rube) is that Notion AI seems to be more oriented towards collaborative writing. Something akin to a learning management software meets WordPress. It’s also a bit more accessible straight out of the gate (again, I’m a rube). This could mean that your students may try/encounter Notion AI before ChatGPT, or that if they try AI writing, they might have an easier time navigating Notion AI. One of your students actually put me onto Notion…