A friend showed me this hilarious website, https://anycrap.shop, where you can prompt for a fake product that doesn't exist and get back a real looking shop page for it. Your imagination can run wild 😆.

I just happened so happen to get ideas for fake products all the time. However, my prompt did not quite produce the image I hoped for, but I tried my prompt on Chat GPT as well for kicks ..
 

of course my prompt was more extensive but you get the idea hah.

I love this website https://anycrap.shop, though, because it seems to symbolize that we are still in the super early stages of gen ai. You can use it to make a lot of fake, well, crap, that no one can buy 😆.

Before online shopping took off, the internet was mostly blogs but also a giant virtual wall to throw all your ideas and see which one sticks.

Writing code was the first most obvious target of gen ai since there is plenty of open source code out there not bound by annoying copyrights, but for now metr.org and maybe others show that AI tooling can make developing slower since you are doing damage control on the long tail bugs in the code .

But anyways writing code was never really the slow part in real world systems. Debugging, troubleshooting, iterating, integrating/gluing, migrating, communicating and other activities where attention to detail is important are among the activities that are hard. 

And I recall in the pragmatic programmer interview with Laura Tacho, from  https://getdx.com, that deciphering stack traces , is the kind of surprising activity where new AI tooling is really shining  .  ( Personally I can confirm this benefit as well !)