Mix & Move with AI

The Mix & Move with AI tutorial for Hour of AI gives students the opportunity to explore AI through image generation and code generation.  In both cases, the student constructs a prompt sentence using a fixed template and a fixed set of options.  

We generated this content ahead of time and stored it on our server, allowing us to review it and to handle high usage during the Hour of AI.  The tutorial retrieves this pre-cached content from our server. Each time, it randomly picks from a collection of variants for a given sentence prompt to deliver a simulation of the variety that is typical of AI generation.  The tutorial also simulates a delay, which is typical of AI generation, when retrieving this content.

For images, the student is exploring content generated using the GPT Image 1 model by OpenAI, rendered as 2D bitmap images as the heads of dancers.  There are over 4,000 pre-generated images over three separate prompt sentences.  The images were reviewed individually for appropriateness and technical issues, and, in some cases, manually adjusted or regenerated.  An example of a technical problem was eyes or teeth that had transparent pixels.  The dancers' bodies are human-created art, with their colors adjusted when loaded to complement the colors in the head image.

For code, the student is exploring content generated on Gemini Flash 2.5 by Google, and presented as block-based code that plays a piece of music using Music Lab.  There are over 32,000 pre-generated code sequences that play music over two separate prompt sentences.  The code is generated as simple pseudo-code, which is then transformed at runtime into block-based code.  Software validation was used to ensure that the generated pseudo-code was valid.