Thinking with AI vs "Problem Solving"

Let me preface this by saying I use AI everyday at work, I write and summarize the emails I write with AI, I use AI to generate faux data for UI elements (instead of using Lorem ipsum) and I also generate images with AI (like the one in this article). I constantly try out new AI tools for potential whenever a new AI tool reach some hype on Twitter, Threads, Reddit or ProductHunt.

But the most I use AI by far is for coding, I generate new files with AI from very specific prompts, I ask the AI what's wrong with my code whenever I get an error or an unexpected result and I prompt the AI for new features all the time.

I mostly use Open AI’s ChatGPT for that and sometimes Microsoft's Co-pilot, as it's built right into Visual Studio Code- my IDE of choice. I tried Cursor for a while to replace it but to be blunt I didn't find its AI to be any better for my usage- so I stayed with VS Code.

Having said all that, that's just for my work projects. In my own personal projects I don't use much AI at all, let me explain why and how I do use it when needed.

After saying all that, you might expect I’d use AI in all my projects but in my own personal projects I tend to not use it as much. When I make a CodePen or a new GitHub repo for myself- I do it to learn and to have fun, so everything I make in my free time is hand coded.

I want to remind myself that writing code is fun and the point of the project is to learn and to have fun- if I use AI to write my code I won’t learn anything and I’ll have less fun.

I do the same with writing, in work emails I might ask the AI to edit it but here in my blog I won’t. Everything you read here are my own words, because the point of writing for fun (which is what this blog is) is to write and have fun, not to publish what the AI wrote- actually write.

Of course I sometimes get stuck so I consult the AI like I would talk to a friend about a specific issue, but never with the attention of making the AI write it for me, just thinking with me.

At this point I have to separate the two modes of working with AI, what I do for work is “problem solving” with AI. Problem solving, which is basically solution hunting, it’s what I used to do with Stack Overflow before AI was even available.

It is to solve my problem as fast as possible, I don’t really read any documentation the AI gives me, I just ask it for code and then I copy said code for using in my project. It’s as simple as that.

It really lands itself to the workplace as I need to find the solution as fast as possible and I don’t really care about elegance or the challenge of writing it myself and all that.

In my personal projects- I do care about the code itself, not only I want it to be elegant and coherent, I also want to write most of it myself because the fun in writing code is- writing the code.

So I sometimes think with the AI as I would talk to a friend about my problem, not to find the solution but to understand my own problem better through talking it out. You can think of it as talking about your code to a non-dev friend, they can’t really help you but they can shed a different light on it so you can find the solution yourself.

I write this blog myself, I never ask any AI to write it for me and I believe I’ll never will. I wanted to write this blog because writing helps me think- so why would I outsource the writing part?

Having said that I do sometimes think with the AI when writing, much like I do when I write code for my side projects. While writing “thinking with AI” is a bit different, as every fact needs to be checked and verified and never taken at face value.

Also, much like a friend who will just make up stuff they don’t know or not sure about. But still the AI can help me think more deeply sometimes. Most of the added value of it comes from its conversational nature because it really does feel like a dialogue between me and whoever I need it to be on the other side.

In conversation sometimes it’s easier to see the holes in my arguments and so it does inform my writing, even if no word or sentence were copied and pasted back to the post itself.

Thinking through a problem using an AI as a tool isn’t new or revolutionary but it is the best way in my opinion to use it, as it deepens the thinking and makes your arguments stronger.

The AI starts being problematic when it’s used as the only tool in your arsenal, without fact checking and critical thinking the AI isn’t a tool, it’s a crutch. When using AI for solution hunting it isn’t a bad tool to find solutions, but it does not sharpen your thinking, maybe it even has the opposite effect.

Whatever it is that you love to do, just make sure you find a place to keep doing it and don’t fall into the hype that says everything should be done with AI. AI is just another tool and we should make sure we’re not using it as the only tool, otherwise we’re not sharpening our thinking.

Think of it as a 3D animator drawing in his sketchbook in his spare time, of course he makes money from his 3D work, but in his spare time he’s drawing like he’d done since he was a child, no fancy tools and no technology to upgrade.

It might sound mundane but that’s what he loves doing, just like I love writing simple code myself with no AI.

Just like I did when I was a child.