AI is a tricky topic for designers. On the one hand, automating tedious tasks such as removing a background and unwanted objects in imagery, or expanding an image are huge time-savers. On the other hand, models were trained on works they didn’t have permission to use. Original creative works carry copyright by default for a reason. Recently, someone described the AI business model as a school bake sale – individuals buy the ingredients and make the baked goods, while the school receives all the proceeds, and the school is for-profit.
Despite recent research showing 60-94% error rates on returned information, AI is here to stay. It will get better (or go off the deep end and we’ll soon bow to our robot overlord, Hal). And companies, like Adobe, are extremely aware of copyright, only training their models on materials they have the rights to use.
A couple years ago, I published a post on the dos and don’ts of choosing solar stock photography. If you’re choosing stock photography, all eleven pieces of advice still hold true. One of the more popular posts in the Creative Purpose blog, it got me thinking about how we can better showcase solar and clean energy technologies using AI in creative. With more and more clean energy companies relying on AI for content and imagery, how can we ensure what we’re putting out there is the best it can be?
Solar, energy storage, and clean energy technologies have come a long way in just a few years. AI tools seem to be on an even faster track. Read on to learn more about what to look out for when using AI in your company’s creative process.
Dos and Don’ts
Image resolution
Most images generated by AI are fine for online use. Images will likely work in print collateral. However, don’t expect to use that image as a backdrop in your booth at the next clean energy tradeshow. Even going through enhancements, images will be pixelated. Know the limits of the generator you’re using.
Nature abhors straight lines
AI is getting so much better at generating solar panels. However, pay particular attention to the background panels and panels at the horizon line. AI still seems to have a difficult time with straight lines of panels in the distance.
Remember copyright*
Some ai models, as noted earlier, have been trained using materials the company has secured the rights to use. Others have not. Be clear in what source you’re using. If your company doesn’t yet have standards for ethical ai use, consider adopting a set of standards.
Also know that copyrighting ai-generated content is incredibly grey at the moment. Logos may not be trademarkable.
And finally, pulling a photo or illustration off the internet and using it as a model to create your own image is – at best – a grey area. By default, unless a creator releases an original work as creative commons, they hold the copyright. Use images marked as creative commons, stock images you have licensed the rights to, or your own/company’s images as source material.
Evaluating Results – Just the Basics
AI doesn’t actually do “new”. AI results are based on an amalgamation of what has been produced before. And for the most part, what gets created is fine for that quick social post or a website mockup to present an idea. However, a lot of bad design exists out there and is being fed into the models. Without getting a degree in the psychology of design, understanding – at a very high level – a few tenants will help you evaluate the results generated by AI to determine if it’s helping or hurting your clean energy brand.
Color
Color is complex. Colors carry meaning. The meaning changes depending on culture. Colors complement each other. Colors clash. Colors have hues and tints. Brightness and saturation. Most brand generators out there ask you to choose your favorite colors. This may not be the best approach to convey the meaning you’d like for your brand. Take a bit of time to understand color meaning in your target audience’s culture before diving head-first into an ai-generated color palette.
Hierarchy
For most, the focus of our attention on any given piece is guided by a hierarchical structure. Assuming for a moment that what we’re viewing isn’t a crowded mish-mash, we will find the largest image first, then the largest text, and then work down from there to smaller text and imagery. Because human brains are hard-wired to look for human faces, the exception to this pattern may come in if there is a single human image in the mix. Our eyes will be drawn to that first.
Choose one item – text, headline, image – that is the most important. That will be your anchor and should be the largest/boldest/most vibrant. Everything else should roll down from that size-wise. For text, keep in mind 3:4 as a general guide. Text should be at least 25% smaller than the main headline/message.
Grouping
We tend to naturally group visual elements that are similar, close to each other, or in a noticeable pattern together. Nine squares aligned in a 3 x 3 grid are more likely to be processed as a single pattern in our minds. Called Gestalt Principles, we can the principles to our advantage by grouping like with like, lowering the viewer’s cognitive load. Alternatively, we can ensure one item stands out by grouping it with other unlike items. If our nine-square grid is made up of grey squares and one yellow square, the yellow square will stand out even more than it would on its own.
Remember your audience
AI is notorious for stereotyping. Confirmation bias is alive and well with our robot overlords. You’ve spent time developing audience archetypes – use them. Imagery should reflect your actual target audience, not what AI “thinks” your audience looks like. Clear prompts will help with this.
Accessibility
Good contrast, skipping color pairings that are common in color blindness, alt-text for images, and good structure for screen readers are all low-hanging fruit to ensure your images, illustrations, and websites are accessible to all viewers. Include instructions to comply with WCAG and ADA standards when prompting AI. And you can always use an accessibility checker like this one or this one.
There is so much more that goes into every design and image created by good creatives. I joke that it is a complex system of trusses, pulleys, and levers. Most of the time, you won’t even notice that the above plus a myriad of other principles and practices have gone into what you’re looking at when presented with a design. AI doesn’t always know either – only what it has been fed. Understanding the basics and incorporating those principles into your AI prompting and use, will go a long way toward ensuring your clean energy brand is represented well.
If you are thinking about how your clean energy brand is being represented by your visuals and marketing, but don’t know where to start in your evaluation, reach out to discuss a design audit.
*I’m not an attorney, nor do I play one on T.V.