I never used to be a person who coded on the weekends.
Not because I didn't have ideas, but because it felt unfair to my partner to sit holed up at my desk hacking away all weekend instead of spending time with him.
Coding with AI tooling has enabled me to scratch the itch of creating silly things with no business value just for fun - and be able to share that experience because the barrier to add value is just natural language and now creating a program can be a fun game or puzzle we do together.
This weekend I wanted to recreate the viral speed reading video as an application that pulled in unique articles but also tests you on reading comprehension. I wanted to do this because after I took that test I felt like I was the fastest reader ever! But no clue if I knew what I just read.
Since I wanted to do this with my husband and he has zero technical background (or interest), I started with goose desktop. Together we crafted the initial prompt around the idea.
This was a very different way of working for me. No real plan, I didn't specify the tech. It was about having fun and making something together.
We started very simple...
...with this prompt.
I want to make a website based on the viral speed reading test. This site should pick a wikipedia article at random for the test and have an input at the end where the player has to put in what they understood about what they just read. The site should then be able to judge the reading comprehension. Help me design a plan to implement this. The first step will be understanding what the viral speed reading test is so you will need to research that
We used Sonnet 4.5 as there was no reason to call on anything more robust here. Not surprisingly, the model chose to create a NextJS application. React is king - no argument from me!
Once my husband saw it working, he had ideas! He is a comms professional with a background in journalism so immediately wanted the font to be more readable and fix that the red letter focal point so it wouldn't bounce around as much.
He was pointing at the screen and telling me what to tell goose to do. We iterated a bit together like this and ended up with a great starting point.
I wanted to add a way to pull a fresh article and a loading state in case the first article returned wasn't long enough. So I added that next.
Here's the test and restart flow.
While this was truly a for fun exercise, it was also interesting for me to see how someone very far removed from any tech understanding use a tool like this and how he problem solved. He was more likely to describe how he wanted the application to feel or focus on the design. Very different from the developers I focus on for work!
As a developer, it was definitely hard for me to walk away from the "finished" project. I could see the rough edges and started thinking about user experience issues - what happens if there's network issues, what happens if it takes 20 tries to find something long enough, how should it handle duplicate content, does it work on mobile....on and on...
At the end of the day though, this wasn't about making a polished product. This was about a new world of fun that you can share with anyone, even your most non-technical friend or partner. Perhaps the future of social interactions includes ephemeral games we spin up and play over an evening together.
There is a lot of hate in the world for AI and lots of valid points to explore there but in many ways this technology can bring us closer to people we love and find new ways to connect with them.
Top comments (0)