5 ways we’ve used AI in our creative practice this year and what we learned.
AI technology is growing in leaps and bounds and we’ve been experimenting with it in our creative practice. It’s become extremely useful in areas that we hadn’t planned on, and not as useful in areas where we thought it should be a slam dunk. Here’s a run-down of 5 different ways we’ve been using AI and what works (hot) and what left us wanting (bot).
1) Making images
In March Adobe launched Firefly, a generative AI tool that creates images from scratch with a few written prompts. The powerful AI turns out photo-real images in just a few seconds. You’d think that this would quickly put a lot of photographers out of business, but it looks like we’re a ways off from that yet.
Hot ?
Storyboards
What Firefly is great at is turning out a lot of images quickly. We were able to successfully use it to generate storyboards for a live action shoot. It was able to help us quickly dial in framing and poses that were used in the final boards and then use talented humans to bring it to life.
Rapid Iteration
The other superpower of AI for images is the ability to take an image idea and come up with different options very quickly. The final execution still has to be done by a human, but AI can help us get further faster when working through options for a visual concept.
Bot ?
Hot dog fingers
Anytime you generate an image of a person where hands are shown, you get some weird hand shapes that look like hot dog fingers. Often this makes an otherwise perfect image, unusable and in need of some serious Photoshopping.
Dead eyes
Photo real portraits look fantastic, until you look into the eyes. They are somehow always dead in only a way that something devoid of human touch could make. It’s a powerful tool, but there is no way you would want to see an AI photo of a dead-eyed person in an ad, unless you’re going for something creepy.
2) Concept development
Hot ?
Something to push against
AI outputs are literally the average of what’s out there on the internet. Using AI and its average suggestions can be useful if you take them as something to push against. It’s akin to a thought experiment; if this is what is out there, what isn’t out there by comparison? AI is great at identifying the lowest hanging fruit, the kind of stuff that we eliminate early in the creative process. So, if you use AI to generate some ideas to poke holes in or turn inside out, you can end up in new creative territories with a little less pain in getting there.
Bot ?
Weak or weird concepts
In a dream world you could feed a well-written brief into an AI and have brilliant creative come out the other end. The truth is that AI is not great at developing an original creative idea. You usually end up with either something very obvious or something that makes almost no sense.
Who can forget the AI generated Blue Jays commercial that made the rounds on social media this summer (warning, this can’t be unseen). This is where AI shows its true colours, in that it completely lacks the context of human experience. This is really at odds with the work of developing original ideas that break through.
Unintended plagiarism
As mentioned above, AI is an average of what’s available on the internet. In a business where intellectual property is important, you can’t be assured that your AI isn’t just lifting someone else’s idea. At best, it will be a copycat idea, at worst you could get sued. Also, IP generated by AI is still ineligible for copyright protection. This is a showstopper for any kind of brand campaign.
Time vs. output
AI can become a monumental rabbit hole. We set a time limit on how long we’ll allow for experimentation with AI during our process. The question we always ask is – could you get the job done faster by just sitting down and doing the work? AI is like going to the casino for output. You may get lucky in the first few pulls of the slot machine, or you could be sitting there all night trying again and again.
3) Writing
Hot ?
Popular topics generation
If you’re looking for a way to get a list of popular topics to get your creative juices going – ChatGPT is super helpful. It can spit these out pretty quickly and you can use the popular topic suggestions as a start for content that you can put your own twist on. This is the classic “further faster” use case for AI.
Bot ?
Original Content writing
Again, the law of averages catches up with you. Your content will be very similar to a lot of others out there on the internet, with the added risk of AI introducing plagiarism to your work. Add in the risk that you may unknowingly be picking up on other brand language and we’re headed straight for some murky territory.
Media releases
Meltwater introduced an AI powered tool called “The PR Assistant” that uses ChatGPT to help create a “high-quality first draft”. This could be great as an iterative tool, but we’re skeptical about its ability to develop a truly unique hook based on our experience with ChatGPT and concept development. Add in the complexity around each media outlet’s AI policy and it starts to get tricky. For example, the Globe and Mail’s Code of Conduct states: “It is unacceptable to represent another person’s [or machine’s] work as your own.” So, by rights you should be flagging if your release was either partially or fully generated by AI when you are pitching it to a media outlet. That would make your work seem less credible in an environment where pitching is already hard enough.
4) Sifting and sorting
Hot ?
We had a video project this summer where we had to do 30-minute interviews with leading medical experts from around the world and condense them down into one-to-two-minute vignettes. We experimented by taking transcripts of the interviews and then had ChatGPT summarize the key points into a short format. This allowed us to quickly paper edit the content and write intros and outros for their clips. Since then, Adobe Premiere has come out with a text to edit function that allows you to do this directly inside the video editing software. Nothing replaces interview prep and skill, but this technique helped us save at least one or two rounds of editing and get to a better product more quickly.
5) Writer’s block
Hot ?
We’re always working on deadlines and as part of that work, writer’s block can sometimes get in the way. We’ve found that AI can be that first push to get momentum. How it gets things moving can take many forms. It could be that there is one nugget in the AI draft that gets you thinking, or that the AI draft is so bad, it makes you mad and motivates you to do better than ChatGPT. At any rate, it can be something that gets you around Writer’s block in a way that wasn’t available before.
The Bot?tom line:
AI shows a lot of promise, but in the creative field, it’s not ready to be the star of the show just yet. Good creative still takes originality and a human touch to resonate with other humans. AI can iterate, get bad ideas out of the way and take you further faster when there is a mountain of work to do.
American painter Chuck Close once said “Inspiration is for amateurs; the rest of us just show up and get to work.” We’ve found that AI can help you do just that. It makes showing up and doing the work just a little bit easier, especially on days where you can use a little extra push.
– This post was completely written by a human ?