How I Learned to Code without Coding: Using ChatBots to build AI, Apps, Websites ...

Never touched a line of code? - but still want to try programming -> this article is for you!

Don’t worry, most of the code will be generated for you.

My journey has been a steep learning curve, and the more I use these techniques, the better I get. What makes programming with ChatGPT so pleasing is that you can create basic prototypes of your ideas in just a day, even without prior coding experience. The ability to see your ideas come to life quickly is incredibly motivating. While becoming an expert takes time and practice, this approach is less about perfecting a tool and more about creatively experimenting and learning by doing.

Use ChatGPT and ask plenty of Questions

ChatGPT acts like a collaborative coding partner and an expert you can consult about anything. It can guide you step by step through the process of starting your coding project. While ChatGPT works well, other AI models can also handle the tasks involved.

• Begin by getting a basic understanding of the algorithm you’ll use. This includes choosing the right coding language, identifying existing code (frameworks), and understanding relevant models or tools. Don’t hesitate to ask for detailed explanations until everything makes sense to you. No code yet!

Here’s a tip: Open a new chat when you want to dive into a new topic or idea. ChatGPT remembers the context of your current chat, which is helpful for follow-up questions.

• Algorithm outline

With the help of your favourite chatbot, sketch out all the necessary steps your project requires. This outline will break down the larger algorithm into smaller, more manageable tasks, which you can then tackle one by one when generating code.

Write and Run Your Code: Google Colab

Google Colaboratory (https://colab.google) is an excellent platform for coding because you don’t need to install anything on your computer. It provides access to free, powerful online GPUs, which makes it especially useful for tasks like data analysis and machine learning.

In Colab, you write your code directly into cells, and each cell can be executed independently. This is particularly helpful for breaking down your algorithm outline into smaller, manageable steps that can be written and tested in its own cell.

If you’re new to Colab, here’s a beginner-friendly guide to get started: Getting Started with Google Colab. The interface is intuitive, so you can also dive in and experiment directly.

For inspiration, check out these great resources:

• Google Colaboratory Sample Notebooks
• Awesome Colab Notebooks (GitHub)

Many of these notebooks feature projects of varying complexity, from simple examples to advanced AI applications. If you’re just starting, focus on the simpler examples before moving to more complex ones.

In the beginning, it’s a good idea to stick with Colab for developing prototypes. As your projects grow more serious or complex, you can always transition to using an IDE on your computer for greater flexibility and control.

How to code and correct Errors

• Start by tackling the first step of your Algorithm Outline. Use a clear prompt to guide ChatGPT in generating the first cell of your code. Make sure to specify that you’re using Google Colab and coding with cells. Don’t worry if you don’t understand the code at first—you’ll gradually learn by reading it as you go along.
• Breaking your project into manageable packages using the outline ensures that neither you nor ChatGPT becomes overwhelmed. Stick to the outline and use targeted prompts like, “Help me write the next cell for [XY step].”

If the code runs without an error message, that’s a great first step! Keep in mind, though, that this doesn’t necessarily mean the code is doing exactly what you want. To understand what it’s doing, examine the output carefully. ChatGPT often includes print("XY") statements to show what’s happening in the algorithm, which can help you follow along.

• Handling Errors: This is where I learned the most about understanding how the code works. Reading error messages and asking ChatGPT to explain them provided some of the deepest insights into what was actually happening.

Even professional programmers spend a significant amount of time "debugging", which often means fixing or adapting code they’ve found online to fit their specific project. Embrace the process of troubleshooting as an essential part of learning and improving your coding skills.

At some point, you may encounter the boundaries of what’s feasible within Google Colaboratory. While it’s an excellent platform for tasks like data analysis and machine learning, thanks to its free access to GPUs and pre-installed packages, some projects may eventually outgrow its capabilities.

For example, I started many of my projects in Colab, including my research paper, "Algorithmic Muse: Robotic-ArmLearns to Draw Humans". After several months of development in Colab, I transitioned the project to my personal computer to connect it with a physical robot. This effort became my artistic installation, "ACV Project".

I hope this guide has inspired you to dive into your first coding project and helped ease any fears or doubts about where and how to begin.

My Background:

At the end of 2023, I began my interdisciplinary Master's degree, which challenged me to explore and apply methods from fields I had never previously encountered. With an architect’s mindset from my Bachelor's studies, I was fascinated by the opportunity to approach design problems from entirely new perspectives, such as those of a biologist or computer scientist.

This curiosity soon led me into new areas: computer science, and more specifically, artificial intelligence with a focus on reinforcement learning.

My Master: Design & Computation at the Technische Universit?t Berlin, Universit?t der Künste Berlin and the research platform New Practice in Art and Technology