Now that generative AI tools are widely available, creatives are using them to spark new ideas, get unstuck, or even just get started. 

Here are the best ways I've discovered to use AI tools in my creative work.

1. Solving the blank page problem

We've all been in the situation where starting work on a creative project just seems so hard. But AI tools can help kickstart the creative process by lessening the cognitive load of beginning work – so you can stop staring at that daunting blank page.

I think about this as changing the difficult creative task of conceiving something from scratch to choosing something from a range of options that you can then alter as you see fit. You might ask ChatGPT to create a bad first draft that you can edit to oblivion or ask it to summarize your messy notes about the project you're working on. Sometimes even having an AI tool create an initial template to fill in is helpful.

The nice thing is that AI tools don't have to get it right, either. When I've used this technique, I found that even when the output is completely wrong, it can help clarify my thoughts by showing me what my idea is not.‍

2. Rubber ducking

Software developers often use a technique called "rubber ducking" to untangle tough problems. The idea is to explain your problem to an inanimate object like a rubber duck. Even though the rubber duck doesn’t answer back, by saying your challenges out loud, you gain fresh insights.

With conversational AI tools, though, the rubber duck can quack back — and potentially give you even deeper insight into your problem. But remember that rubber ducking hinges on you unraveling the issue. While ChatGPT and similar AIs can propose solutions and ask insightful questions, you're still the one at the helm — and often the act of typing the prompts is more valuable than the response you get back.

This technique is especially good for fiction writers. By using a clever prompting set up, you can ask ChatGPT to have a conversation with you as if they were your character. You can then ask them for help working through their story.

3. Filling in the details

In my creative work, I often find myself needing to come up with a specific detail that doesn't make a real difference in the end product, but has to "feel right" in some abstract way. I'm talking about things like naming a minor character or the town they grew up in, or nailing a descriptive detail for a minor plot point. These snags can disrupt your creative flow because they take a lot of time, but the prose doesn't come together in a polished way without them.

Using AI tools to come up with ideas and using your judgment to narrow it back down to something that fits is a way to quickly solve this problem — even if you end up using it as a placeholder and changing it later. Text generation tools can help with coming up with names and examples, and image generation tools can give you some ideas for details of a place or person.

Don’t be shy about asking the AI for more examples, since a longer list of possibilities gives you a larger canvas to experiment. The trick is to make sure you're clear on the parameters of what you're asking — otherwise you might never get to where you want to be.

4. Thinking outside the box

Bob Ross is famous for saying “we don’t make mistakes, just happy little accidents.” This also applies to AI tools, which can generate an unexpected element that helps push your idea forward.

Image generation tools are a great way to find these accidents. Specifying something vague, abstract, or strange can allow you to deliberately generate something odd to mine for ideas. Some creatives swear by using nonsensical prompts.

For one of my creative projects, I was trying to get a better sense of one of the characters. I had a strong sense of the character's part in the story, but didn’t have as many details about her personality or anything distinctive about her appearance. To give myself some ideas, I fed Midjourney an abstract, vague description. After several attempts, it returned with a pencil drawing that perfectly depicted the character. An unexpected but pleasant surprise was the presence of a second character that inspired me to incorporate a younger sister into the storyline.

But you don't have to generate something strange for randomization to benefit your creative process. Even prompting generalities can give you some ideas to work from.

5. Encouraging lateral thinking

Lateral thinking, a term coined by psychologist Edward de Bono, is a creative problem-solving technique that involves looking at a situation or problem from different angles rather than from one set perspective. It can help generate innovative solutions and ideas that may not be immediately obvious.

A great example of lateral thinking in the creative field is the work of celebrated musician and producer Brian Eno. Eno's inventive approach to composition led him to create the Oblique Strategies card deck where each card has a cryptic instruction or constraint intended to break creative deadlock and encourage innovative thinking. The actual deck is pretty extensive, but with AI, you can tailor the idea to your specific project or problem.

When I used this prompt in ChatGPT "Imagine that I've drawn a card from the Oblique Strategies deck. What does it say?", it replied "Focus on the Flaws." 

6. Focus on the flaws

Most of the time we want the AI tools to work better. Yet, there's an unexpected source of inspiration that we might overlook: the flaws within the AI tools themselves. 

A few months ago, I was working on some creative concepts for a science fiction story using an early version of Midjourney. I asked it to render an image of a wine bottle and rather than a relatively straightforward depiction, it created one that looked like it was covered by an impressionist fall forest.

Rather than discarding this “mistake,” I found myself fascinated by the unexpected output. It gave me a concept — that the characters would decorate and reuse glass wine bottles — that fit so perfectly with the society I'd imagined that I added it to the story.

Likewise, old language models sometimes get stuck in odd loops or otherwise generate useful gibberish. I've named a character inspired by a nonsensical repetition I got from Google’s T5 language model from 2020.

This is a great example of how human creativity is still in the driver’s seat when using AI tools. No AI tool will tell you how to repurpose its mistakes; you need to be the one to link these random possibilities to your work. Often mistakes are just mistakes, but sometimes you'll land on something great.

7. Rounding out your ideas

In my creative workflow, I’ll usually create an outline without using AI tools and then ask ChatGPT to complete the same task. I find this helps me round out my outline — sometimes it suggests ideas that I hadn't thought of, which lets me decide if I want to include them in the piece. But I think its real strength is in challenging my thought process. There have been several times where it has justified an idea I rejected too quickly, which let me reevaluate whether I wanted to backtrack on my initial decision.

Case in point: For the article you’re reading now, “creating multimedia mood boards” (below) is a mashup of two of ChatGPT's ideas (soundtrack creation and mood boards) that I didn't feel were good enough to include in the article on their own. Combined, though, I felt they added a missing angle to the article.

There are a few ways you can have ChatGPT round out your ideas. You can prompt it for more examples, ask it to focus its output on a certain area, or request that it flesh out specific ideas.  

8. Building your AI skills through creative play

This one’s kind of meta, since the creative exercise ends up making you better at using AI. Some of my favorite channels on the Midjourney Discord are those that have prompt games where players share, rate, or otherwise use generated images as part of creative play. It's not just fun, it also gives you practice prompting so that you can get better. It can extend the range of your capabilities with the tools by pushing you to generate images in styles you might not otherwise have tried.

You might find (or generate!) a list of creative prompts to follow, or enlist a partner or group to play.

My favorite prompt games are Jewels in the Sand, an adaptation of a board game where players try to guess the rule that separates "Jewel" images from "Sand" images by submitting images they generate and guessing the rule; Telephone, where players try to create an image as close as possible to the one they receive; and collaborative storytelling where players take turns telling a visual and written story.

9. Creating multimedia mood boards

As the composer for the movie Dune, Hans Zimmer didn’t just write the soundtrack. He also produced “The Dune Sketchbook,” which included extended musical explorations of the original score. The Sketchbook tracks weren’t included in the movie itself, but were written to immerse listeners in the mood and surroundings of the story.

Chances are you’re not a Hans Zimmer-caliber composer. But AI music generation tools like Amper Music and AIVA still give you a way to create musical snippets that can be used to get into the mindset of your creative world. Likewise, AI image generation tools like Midjourney and DALL-E can be used to visualize and conceptualize the visual style and mood of your work. You can even use AI voices — like Descript’s — to narrate parts of your dialogue, so you get a feel for how it might sound out loud.

I've found a lot of success making mashups of music, videos, and images that combine specific elements I envision for my story but don't quite match anything already out there.

10. Giving you confidence in the ideas you already have

If you’ve ever flipped a coin to decide between two seemingly equal options, watched it land on “heads,” and immediately thought “Oh man, I wish it had landed on tails,” you know how an outside influence can help you realize what you really want. 

Sometimes, the AI fails. Sometimes, none of the ideas ChatGPT comes up with are superior to the ones in your head. That in and of itself tells you something about your ideas: you actually think they’re pretty good. In that case, wish the robots well and go forward with your own human creativity.