Augmented Development with AI

May 17, 2023

I recently gave a talk at a local meetup about using AIs to help you write code. I knew basically nothing about the topic when I volunteered and I learned a ton along the way. This post is a brief summary of what I shared there.

The main ideas

If you feel apprehensive about AI, get curious! - We tend to fear the unknown so start learning! Read stuff. Sign up for AI tools like ChatGPT. Try building something with an AI assistant. Most importantly, find a reason to get excited about AI. For me, it was the idea that AI could help me build the little scripts and side-projects I had previously dismissed as not being worth the effort (like this one and this one). There are many possible benefits to AI (making your work less boring, making you more productive, rapid prototyping, expanding the breadth of your capabilities, etc). Find the benefit that speaks to you.

It’s easiest to start by using side-projects - AI tools for coding are very new and may not be ready for use in the workplace. Some employers have prohibited them, due to concerns around data leaks. Even if they are allowed, the current tools likely don’t understand the full context of your large enterprise codebase when making suggestions. Using AI for simpler, less risky side-projects gives you a safer place to learn and get some easy wins.

It will take practice to get good at using AIs - When search engines were still pretty new, there used to be all these articles about “google search tricks” to help you find what you wanted. We’re in a similar phase of AI usage right now. As you use AIs for coding, you start to get a feel for the things they do well and things they don’t. You also get better at prompting the AI for what you want. Now is the best time to practice these skills, while the AIs are free or inexpensive to use.

The specific tools will come and go - Some popular tools for coding assistance today include ChatGPT, Github Copilot, Phind, and Cursor. The space is rapidly changing though. Instead of tying yourself to one specific tool, be willing to jump around and try a lot of things.

My AI reading list

My talk also included an “AI reading list” with all my favorite articles on the topic:

If you know of any good ones I didn’t mention email them to me or leave them in the comments.

The full talk

If you’re interested in seeing the full talk, here’s a recording the organizers put together.

Comments