The thing about GPT-4 that most feels like a superpower is that I can work with it to write programs to automate annoying tasks.
I have already co-created two Python programs in the last week to solve little problems, and a Unity program for fun.
I don't know these languages.
I have already co-created two Python programs in the last week to solve little problems, and a Unity program for fun.
I don't know these languages.
When I encounter errors or the output isn't what I like, I just tell GPT-4 and it fixes the problem and explains what was wrong.
It is like giving someone instructions to build software, but it is done instantly and you don't feel bad asking for as many corrections as you need.
It is like giving someone instructions to build software, but it is done instantly and you don't feel bad asking for as many corrections as you need.
I should mention that, just like any other application in which I use AI, you need to understand the outputs, if not the process.
I wouldn't trust it if I couldn't check the results myself to ensure that it was doing the right thing. This is a force multiplier, not replacement.
I wouldn't trust it if I couldn't check the results myself to ensure that it was doing the right thing. This is a force multiplier, not replacement.
"I need to create an Amazon Echo skill that will flash my hue lights green and blue when I yell party. Can you create it?"
Also, to be clear: longer programs can be much more of a pain to generate this way, because the model starts to forget what it told you earlier. Little scripts (or projects with many pieces) have worked better for me.
And you do usually need to do some debugging, with AI help.
And you do usually need to do some debugging, with AI help.
Reader, it worked.
It was pretty involved to make (I had to set up and configure AWS accounts & Hue APIs, I had no experience with either) but the AI took me though it all, and did all the debugging (75% my errors, 25% AI). But using tutorials would have taken me 10x as long.
It was pretty involved to make (I had to set up and configure AWS accounts & Hue APIs, I had no experience with either) but the AI took me though it all, and did all the debugging (75% my errors, 25% AI). But using tutorials would have taken me 10x as long.
Loading suggestions...