Example of Human-directed AI-enhanced Learning

The way humans use the AI generative image space is becoming more interesting by the day! Here is an example of how this might be applied to education. Below is an image of Midtjourney5 images curated on Notion by TELC from a Twitter Midjourney5 community. Firstly, it is important to emphasise how this kind of community activity can be educational and inherently motivational for would-be student community members, as image prompts are also shared, analysed and developed further by community members. Secondly, it highlights the fascinating interplay of how humans and AIs might work together creatively. Arguably, this is a good example of human-directed AI-enhanced learning. 

In summary, this is the learning process through which the student would go:

Step 1 - The students curate the images with prompts into a library

Here is an example of a library TELC created: Click Here

Step 2 - Students find e.g. five images with their corresponding prompts from the library above they like the most

Tiny Water World

Mechanical Geisha

Astronaut Diorama

Fractal City

Skeleton Businessman

Step 3 - Students train ChatGPT to understand the prompts of their chosen images

Step 4: Students ask ChatGPT to generate 5 new Midjourney prompts based on what it learned, but with the following constraints:

  • Using five different styles (you can ask it to base them on the original promps' styles)
  • With the same subject. (we chose: If Napoleon had lived forever)

Step 5: ChatGPT gives us the new Midjourney prompts

Step 6: Students put the prompts into Midjourney and it generates the images

See The Results!

Step 7: Students study the resulting images and then share them in a new library or to the community at large.

Note: Another interesting activity for students to do is to try and speculate how ChatGPT came up with its prompts, and how altering the prompt it was originally given might affect the images.