Published on

Transforming how we code (and work) with AI

Authors

When I learned to code it was so I could make Flash websites. Interactivity was all the rage and if you knew ActionScript, you could build something cool. Eventually JavaScript took over and I moved my expertise there, building websites and web applications for companies and clients.

I found my ideas always outpaced the speed at which I could them. I did launch a decent number of projects over the years, but when I look at my list of ideas that I’ve wanted to build, in code, what I’ve built is just a small percentage.

I’ve always been good at understanding how things fit together though. I can whip together a product requirements document pretty quick, and I love thinking about information architecture.

With AI I can honestly say that I am now 10x more productive than before. Granted, with years of practice, I do know better how to quickly ask and answer the right questions, then decide what the right thing is to build to take the first step to market.

Coding with AI is not like ordering a pizza from Dominos.. We are not yet at a point where one can tell the AI what to build, then have 100% confidence that it will produce exactly what you have asked for. Sure, it can output code for a simple app with simple interactions, under the right conditions. To build something complex, with many files and functionality, the AI needs someone to direct it, someone who knows the structure and requirements for the project, how the information and interactions flow through it. The right person can ask the AI the right questions so it can succeed, they can examine the output, debug it if needed, and decide the code is good. Coding with AI is like working with a copilot - it is helpful but still needs someone to lead it along the way.

AI tools have only been available to developers since GPT-3’s release in 2020, and we’ve all been using tools like CoPilot, ChatGPT and Perplexity to generate code and more since late 2022. In just under 5 years, simple AI tools are commonplace in many knowledge workplaces, generating boilerplate code, writing polished emails, contributing to brainstorms and more. There will be job displacement and concerns for AI safety are real. While it took several decades for the industrial revolution to transform society, we are in a position where we will see AI drive an even faster and more profound change in just a few years.