PART III — Becoming an Effective AI User
Book 10
Claude: Building Something New
Why I Tried Claude
After using ChatGPT and Gemini, my curiosity shifted from asking questions to creating things. I wanted to build simple educational games for my granddaughter that she could play in a web browser.
As I explored different AI assistants, I repeatedly heard that Claude was especially capable of generating and refining HTML. That immediately caught my attention.
I had been a professional programmer years earlier, so the idea of generating code wasn't intimidating. What intrigued me was something entirely different: could AI dramatically reduce the time it took to turn an idea into a working project?
You don't need a programming background to try this. My earlier programming experience simply gave me confidence to experiment. Claude produced and refined the code through conversation.
What I Discovered
That does not mean Claude simply writes a finished program correctly on the first attempt. Code can contain errors, misunderstand a requirement, or behave differently than expected. The useful part is the conversation: I could describe the problem, test what Claude produced, report what happened, and ask for a revision.
The first version was rarely perfect, but it gave me a starting point. Each conversation improved the project through clarification, testing, and revision.
Instead of writing every line of code myself, I described what I wanted the game to do, asked questions, and refined the result with AI.
That experience reinforced a lesson that applies far beyond programming: AI often works best as a collaborative partner that helps transform ideas into finished projects.
One Project Led to Another
What began as an experiment in creating games became a broader lesson about building with AI. Whether the project involved writing, coding, or organizing ideas, the process was remarkably similar: describe the goal, review the result, refine it, and repeat until it met my needs.
A Little Technical Detail
Claude is a family of generative AI models developed by Anthropic. Like other large AI models, it can work with language and other forms of information, then generate responses based on the instructions and context it receives. One practical capability that mattered in my experiment was generating and revising computer code from a description of what I wanted the program to do.
The appendices provide more technical information about AI models, code generation, and related tools for readers who want to go further.
Try It Today
Think of a small project you have postponed. Ask Claude:
“Help me build this one step at a time. If you need more information, ask me questions before you begin.”