‎There’s a lot of hype around AI tools. Hundreds of tools promise to automate your writing, research, and editing. But as any witness to the failed AI-driven Willy Wonka Experience can attest, there’s a serious gap between AI’s hype and what reasonable creatives can actually use it for.

So we wanted to see how creators are actually implementing AI into their workflows. No over-the-top promises of outsourcing your scripting to a robot, just simple ways to make you a better creative.

Writing with ChatGPT, the right way

Juan Faisal, a creative strategist specializing in generative AI, says he’s figured out ways to productively use ChatGPT in his writing process without falling for its frequent hallucinations. He uses text-to-speech to brainstorm with the platform. 

For an interview podcast he runs, for instance, he feeds ChatGPT the framework for the podcast, explains that he wants a three-act structure, provides information about the person he’s interviewing, and requests a list of questions. Then he’ll choose which questions he’ll continue to refine—only about 30 percent of the questions are usable, he says. 

He also uses it to think through ideas to write about, and will ask the platform to argue for or against a specific position. But he doesn’t just copy and paste its responses. Instead, he uses the back and forth with the chatbot to help him clarify his ideas.

In my own work, I find ChatGPT most helpful to cure writer's block around headlines or introductions. I almost never use its ideas as it presents them, but rewriting an AI-generated introduction is somehow faster than starting from scratch. Even when I don’t use any of its output, its unusable suggestions can spark a better idea.

AI transcription tools

AI-powered transcription tools, such as Descript and Mac Whisper, have been a gamechanger for editing audio and video. While transcription services have been around for a while, they’ve become far cheaper, faster, and more accurate, enabling everyday creators to use them. 

Faisal uses transcripts to speed up his editing. For many creators, the ability to see words on a page helps them visualize the structure of an episode in a way that looking at a timeline in Adobe Audition or Premiere just can’t.

Use transcripts to see your whole project at a high level for major restructuring, or get detailed by cutting down a speaker’s rambly sentences. Transcripts are also invaluable for their searchability. Missing a good piece of tape about a specific point? Transcripts make the process take seconds.

These tools often make mistakes, but those errors don’t get in the way of making good edits. But if you’re looking to use these tools to generate subtitles or captions, it’s important to review them with a human eye.

AI voice enhancers

These are especially invaluable tools for people running remote interviews without the budget to ship their guests good quality microphones, or the technical know-how to clean up noisy audio themselves. Descript pioneered this technology with Studio Sound, and Adobe Voice Enhance can be used for a similar purpose. 

These tools aren’t perfect—the worse the audio is, the more likely you'll get odd sound fragments or artificial-sounding voices. But it’s often a worthwhile tradeoff when professional sound engineering isn’t available.

Ethical image generation: Adobe Firefly

Lots of AI evangelists will use AI-generated imagery with reckless abandon, but many of these tools raise ethical and legal issues. Faisal prefers to use Adobe’s Firefly tool, which is trained on Adobe’s own stock library.

Creator Chris Priestley demonstrates how he can use Generative Fill to add elements to real photographs, such as adding flowers to a wall of vines.

One user on Reddit used it to edit their YouTube thumbnails and remove unwanted elements from images. Another described using it to change an image’s aspect ratio. Generative Fill offers a good middle ground for people who don’t want the hallmark look and feel of AI-generated art, but want to use generative tools to augment their own photographs, thumbnails, and so on.

AI-generated social media clips

Podcasters and other creators of long-form content often struggle to pick what clips they’ll share on platforms like Instagram and Tiktok. In some cases, their own biases may hinder their ability to pick what will actually do well on those platforms.

Many creators are using AI tools to automatically generate shareable clips, though with mixed results. One Redditor used OpusClips to generate social media clips for a documentary, and used the platform’s built-in editor to reframe shots as necessary.

Like other AI tools, it’s important to vet and tweak AI-generated clips. Many might not be usable, but it will still likely save you time.

Descript also has an AI feature called Find Good Clips, which combines the best of AI automation with your own human eye: the AI picks the clips, you polish them so they suit your tastes.

AI voice cloning

AI-generated voices are a mainstay of TikTok, but they still seem very out of place in long-form  and commercial content. Tools that clone your own voice offer another middle ground. Descript’s Overdub is a helpful tool to plug in missing audio or fix mistakes when you’re no longer at the recording studio.

Another use? Scratch audio. It’s often helpful to have placeholder narration as you piece together your footage or tape. While Descript offers traditional text-to-speech generation with stock voices, you can also generate text to speech in your own voice, which may help you better visualize how it’ll sound when you hit the recording booth. TikToker SixteenRamos called it “scary close” to how she sounds conversationally.

A tip for AI use

The fundamental challenge as a writer, filmmaker, or podcaster is to stand out from all the others. Generative AI, when used improperly, will do the exact opposite. It’ll make your content seem exactly the same as every other person who’s writing scripts or generating imagery entirely from tools like ChatGPT. Smart creators should use AI to augment their own talent and knowledge.