We are now about a year and a half into the age of generative AI. There was a lot of fear and anxiety heading into this era. I want to take a moment to assess what the reality of working with this software has been like, from a creative perspective.

Long story short: working with AI is a lot more work than everybody imagined.

Let's start with text.

The major theme of my course about creating content with ChatGPT is that you need to lead. You need to be creative and problem-solve to work around the serious limitations of ChatGPT.

ChatGPT writes generic, dull text, and it's hard for us to read more than a few paragraphs of this stuff without our eyes glazing over. The onus is on you to make that text work.

ChatGPT is a powerful and very worthwhile tool, but the area where it will have the most impact is gruntwork, the boring work you don't wanna do. Actual creative work? It can't do that.

Image generation is even more limited.

Much like there's the "uncanny valley" effect in CGI, AI-generated images feel empty and uninvolving. You can see one of these images at the top of this post. It’s fine for this modest purpose, but for higher-level work, it would seem "temp," like a placeholder. (It also looks like countless other images being pumped out all over the internet.)

But haven't I been using image generation? Yes, I've been experimenting and demonstrating what this software does. But, as I said in my course, I don't think image generation is ready for prime time yet. If AI art is going to work for you, it's your own creativity that will do the heavy lifting.

Text and image generation gives you raw ingredients, much more raw than what you would get from a human collaborator. And it takes a lot of imagination in your preparation to make these ingredients flavorful.