CreativeCArd.png
Posted on by petert

Last week I worked on adding a new lecture to my Creative Concept development course. The lecture focused on using AI in creative work, especially in the early phases of design.

I started with this video


Introduction to a AI and creativity.

And of course the video it is mainly produced by AI tools:

  1. Firstly I created the content using Chat GPT 3.5, “Create a short spoken presentation on how AI can be used in Creative concept development.”
  2. Then the text from Chat GPT and a couple of audio samples of my voice were used to create the synthetic voice using tortoise-tts
  3. Finally, the audio track was used with a still image of me to create the video using the tool D-ID

In the lecture, I explained how AI and machine learning can be used for creative phases in product development, these was the main areas for using AI.

  • Creating ideas with ChatGPT
  • Creating concepts for your ideas with Midjourney and DALL-E
  • Refining sketches that you have done

I demonstrated the tools with a lot of examples and highlighted the importance of writing a precise prompt, and how the prompt can influence the result.


Example of a creative method card generated by Chat GPT and Stable fusion

Example of design sketches created a self watering pot created by Midjourney
Refining a rough sketch using Midjourney
Different interpretations of myself using a start image and refinements in Midjourney.
Changing the prompts for image generation in Midjourney

Exercise 

I also let Chat GPT come up with an exercise and also how to divide students into teams. And Chat GPT responded with a version of the “human Knot” that worked fairly well and come up with the exercise “AI-Assisted Brainstorming.” where students used Chat GPT and DALL-E to come up with ideas to a specific problem.

For the exercise, I had to adjust the output from Chat GPT to make it a bit easier to understand:

  1. Do a quick brainstorming without AI tools (10 min)
  2. Come up with a prompt design to solve your problem (10 min) 
  3. Come up with at least 50 ideas for the problem (10 min)
  4. Select the top three ideas from both ordinary/AI brainstorming (10 min)

The importance of creating a good prompt

Reflection

The lecture went pretty well. I spent to much time on the introduction (about 45 min). Also, it was obvious that it was tough to get good prompts for DALL-E it feels like Midjourney is a bit better on comming up with nice solutions. Some students used Chat GPT to create prompts for DALL-E and that was a winning combination.

I usually use I like…, I wish to gather feedback and this time I used Mentimeter so I, later on, could feed the student’s individual feedback back to Chat GPT to summarise the results.

I like 

The students liked the opportunity to experiment and play with AI during the lecture, and found it helpful for understanding how to use AI in a creative process. They appreciated the practical exercises and found them fun and inspiring. They also enjoyed learning about different AI tools and exploring their possibilities and limitations. They found the lecture to be beneficial in developing their creative strengths and understanding how AI can be used for concept and product development. Some students also appreciated the opportunity to compare AI programs and found the practical application of AI to be useful for solving problems. Overall, the students found the lecture to be beneficial in learning about AI and developing their skills.

I wish…

The students wish they had more variation in the problems they were given during the exercise, and that they had more time to work with both ChatGPT and DALL-E. They also felt that they needed more time to work with the AI tools, and more time to write prompts for the AI. Some students wanted more training in finding good and suitable prompts. They also wanted more time to try the different AI tools and compare them. Some students felt that the introduction to the different tools could have been more detailed, and others wanted more background information on the history and technical aspects of AI. They also suggested that they could have been given more time to practice and play with the programs in order to discover their potential. They also suggested that the group presentation of the pictures could have been shorter. Overall, the students wished they had more time to work with the AI tools and learn more about their potential.

Improvements

I then asked Chat GPT on “how to improve the lecture”

Based on the feedback from the students, here are a few suggestions on how you can improve the lecture:

  • Provide more variation in the problems for the students to work on during the exercise
  • Allow more time for the students to work with both ChatGPT and DALL-E
  • Provide more training and guidance on how to write good prompts for the AI tools
  • Allow more time for the students to try and experiment with the different AI tools, and compare them
  • Provide a more detailed introduction to the different tools and their functions
  • Include more background information on the history and technical aspects of AI
  • Provide more opportunities for the students to practice and play with the programs to discover their potential
  • Reduce the presentation time of pictures
  • Provide more time to work on the solutions a little bit more. 

Reflection on the AI reflection;-)

It is difficult for an AI to improve the lecture based on the “I like…, I wish…” because it needs to have the lecture context and the information on how much time I spent at each phase. To improve the lecture next time, I will provide some material for the students to view before the class and let them use Chat GPT to create prompts for DALL-E or Midjourney.
Teaching AI and creativity