Make It More

Week 3

Competencies: Skilling and Productivity, Art and Creativity, Critical Analysis

Overview

Practice iterative refinement using AI by taking an initial concept or creation and progressively transforming it through specific lenses or directions. Learn to guide AI systems through intentional iteration cycles.

Details

A powerful strategy with chat bots is that you can build on earlier in the ‘conversation’. There was a meme with text-to-image models where users asked it to ‘make it more bro’, and each time the model tried to adapt to the request, in increasingly outlandish ways.

It’s not just a joke-y strategy though. Max Woolf, in ”How to Write Better Code Using AI” (2025), constantly kept asking a chat bot to improve a bit code, and found that it started finding new strategies for optimizing the code.

This is a powerful technique, and it really stems from our willingness to ask followup questions, to ask the system to criticize our work or ideas.

Consider the following instances, repeated ad-nauseum, and instances that they may be applied to both work, ideation, and fun.

• make it simpler
• make it weirder, make it more creative
• make it more practical

Or, consider a second step that I find useful → ask a model to critique on some grounds (e.g. ‘critique this on the grounds of creativity’, then ‘update it to respond to the criticism’).

In practice, you’re not ‘making it more’ forever. However, for this exercise, I want you to observe what happen as you over-optimize. There’s much you can learn, about your material (e.g. if you keep shortening something your wrote, where does the tiny version of your idea end up) and about the model itself.

Choose a starting point (text, idea, code, etc.), either by putting in something of yours or asking a bot to create something to start. The apply progressive refinement through one of these approaches:

1. Perspective Iteration. e.g.
   • Apply different viewpoint critiques
   • Improve based on each critique
   • Example: “Critique as a professor” → “Improve based on critique” → “Critique as an industry expert” → etc.
2. Characteristic Transformation. e.g.
   • Make it simpler/more complex
   • Make it more creative/professional
   • Make it more playful/serious
   • Example: “Make it weirder” → “Make it even weirder” → etc.
3. Target Audience Shift e.g.
   • Transform for different audiences
   • Adapt tone and content appropriately
   • Example: “Explain to a child” → “Make it even simpler” → etc.

You can adapt the ways you ask the bot to iterate in this lab - be creative, no need to use my specific examples! You can even try something avante garde, to see how the model interprets you (e.g. ’make it more Buzzfeed’).

Also try cycling between different types of prompts.

You don’t have to use text outputs, if you’re familiar with multi-modal systems (which haven’t been introduced in class yet).

More Example Sequences:

• Academic paper abstract → Simpler → Even simpler → Child-friendly → Visual metaphor
• Basic code → More efficient → More readable → More elegant → More creative

Completion Details

Try the exercise at least three times, and post the most interesting output.

Post the initial starting point, and examples of how it changed. You can truncate or abridge for clarity, as you see fit. If you have a link to the chat, share that. Add brief commentary on what you observe, as appropriate.

As with our other labs, this is graded only for completion unless you include it in your portfolio. Have fun with it.

Portfolio Details

Size: 1 point

If including in portfolio, just include your post. You’re welcome to add more detail or commentary, but no extra work is necessary.

Grading will focus on: demonstrated understanding of iterative refinement techniques; creative application of different transformation approaches; thoughtful analysis of how the content evolved through iterations.