How this course was made

I deliberately tried to keep this course as "hands off" as possible, knowing that much of the content online will be almost automated in the interests of cost and "efficiency". That being said, I also kept it short (<15 minutes across 3 modules) because I just don't think anyone is ready for an hour of virtual Leon!


In terms of production, I had to make the input video for the AI avatar, prompt ChatGPT for the script, copy/paste the script into HeyGen, and generate/download/edit the video.

The editing was essentially splitting the full video into four sections, adding in-out fade to black transitions, and exporting (I used Camtasia). Finally I had to manually upload to the course platform to the Teachable site, and add this section and the "human" video.

Of all those manual tasks, I imagine that the following could already/soon be automated:

• generate script directly to avatar
• download in already cut segments with transitions
• upload to platform

Additional resources

Although this is an experiment in creating an entirely AI generated course, it also has a serious message about deepfakes and synthetic media.

As educators, we need to explore both the "how" of synthetic media, and the "what", "why", and "what does this mean?"

The course is AI generated, but as you can see in the human intro video I used several genuine inputs as source material to create the script and the slides. Here are some resources either used as input, or useful to the conversation:

• Deepfakes and Higher Education: A Research Agenda and Scoping Review of Synthetic Media is a research agenda from Jasper Roe, Mike Perkins, and me exploring both the potential positives and negatives of synthetic media in education. This article was used as part of the ChatGPT prompt.
• We Need to Talk About Deepfakes this blog post from my blog at leonfurze.com explores the concept of deepfakes and looks particularly at the risks associated with the technology.
• Real or Fake is the AI generated image game mentioned twice in the modules. I created the game on a whim, using Claude to write the code for my website. It went viral, and over 50,000 people have now tried the game. It's very difficult to get a perfect score!
• Digital Plastic: Understanding AI-Generated Synthetic Media is another of my blog posts. It was not used to prepare this course, but provides a good background on the broad topic of synthetic media and my term "digital plastic", which I use to suggest that this media may have both good and bad uses online.
• https://www.esafety.gov.au/ is mentioned throughout the course and my blog posts as an excellent resource for information, and for Australian individuals who need to report image-based abuse including via deepfakes.
• https://takeitdown.ncmec.org/ is a Meta funded organisation that can add digital metadata to images which then results in them being identifiable by search engines, and able to be taken down across the internet.