Exploring Generative AI: A Journey of Learning and Guilt

10.11.23

This is a short description of the blog post.

I love learning and trying new things. Over the past few years, I've felt guilty because I believe I'm not learning and experimenting as much as I'd like. One of the things I feel guilty about is not delving deeper into AI, but that's my own fault as I decided to prioritize family, my health, and my job.

We're all tired of hearing that AI will impact our lives in some way or another. We should all either become AI experts or fear being left behind. At least, that's what it feels like.

Even though I feel guilty, I managed to do a bit of learning and experimenting over the past few years. I read "Life 3.0 - being human in the age of artificial intelligence" by Max Tegmark when it came out in 2017. I started using Dall-E and ChatGPT as soon as they were released. I got Notion AI before it was available to all users. I later subscribed to ChatGPT Plus (but canceled it after feeling like I wasn't getting enough value from it). I have read, watched videos, and attended talks on generative AI. I still feel lost.

Text-to-image generation is one of the areas that I find interesting but haven't experimented much with. The most popular text-to-image generators are probably "Dall-E", "Stable Diffusion", and "Midjourney". I intend to learn more about them in the near future, both how to use them and how to write prompts effectively to get the results I want.

For today, I think that good enough means just creating an image with Dall-E and StableDiffusion XL. I didn't really want to get into Midjourney...

Here is my simple but good enough prompt: "Create a picture of a man who is learning to use generative AI to create pictures so that he can post in a blog post about learning to use generative AI."

And here are the results:

DALL-E

Stable Diffusion

Good enough.