This Is What Happens When You Let Bad AI Run Your Marketing

It was the middle of designing all hands when it came up.

Someone asked:

“Why don’t we just use ChatGPT for our copy revs?”.

Everyone started going nuts for this but as the head of content design, I had to be the bad guy. I wasn’t opposed to AI just because. There was the obvious issue: security risks for proprietary company information. But there was also a thornier issue the designers didn’t get.

In the last ten years, content design has been on the rise, but it hasn’t been the easiest growth. The same trends that make it difficult to do great content design make AI a risk for these organizations. Content designers have been spread thin or struggle to have an influence on our projects.  I’ve commiserated with so many about being brought in at the tail end of a feature, asked to clean up the words, and then have the designer or PM look at you in surprise when it takes more than a day to clean up the flow. Worse, I’ve seen projects go out where engineers accidentally use an old mockup’s copy or even change the content on a whim.

You could argue that these are reasons AI could be helpful, it could help content designers focus on higher level strategy, scale a bit more effectively. But, if content design and strategy lacks systems, checks, and governance, well, AI is only going to exacerbate that problem, not fix it.

Cost #1: Sound Like Everyone Else

We’ve all seen the social media posts: AI has a reputation problem for a lot of consumers. Whether or not people accurately recognize it, people generally don’t like it.

Great example for how negative a lot of people feel about AI. Source: Gina Denny, BlueSky

Creating content with AI, does tend to have a specific voice/sound, unless you are willing to edit it or know how to prompt it with additional cues or examples to emulate.

Cost #2: Increased Inconsistency

Wait, didn’t I just say AI could make us all sound the same? There’s a difference between voice and consistency.

The issue here is that if you lack guidance on how to write specific kinds of content, you can further erode trust with users by creating inconsistent experiences. It can be as simple as using two different words for the same idea. In a social media app, are they contacts or connections? Is it an invite or a request?

On the surface that seems so small, but the issues can grow over time, having features that can overlap or contradict each other. If you aren’t deciding these terms, AI will help you continue to create multiple overlapping terms for the same thing. This can make support that can be incredibly hard to reign in down the road.

Cost #3: Ineffective Content

One of the biggest myths in content design and strategy is that it’s ‘just about the words’. A major part of the work is telling a story through visual layouts, identifying what content makes the most impact where in an experience. AI could evolve to help answer these questions but for now, it’s still limited. For instance, you could have ChatGPT come up with the best way to describe the discount as a header. However, if your user is still worried about the process they're working on, the content is still going to fall flat. Having the ability to question not just what is the right content but where in an experience is still best when you have a human and especially an expert at the helm.

I don’t think we should avoid AI entirely. It can be great for idea generation, or playing with other ways to express an idea. But if you don’t have a solid brand identity or content standards, you’ll start to sound like everyone else. Worse, if you don’t have the right people reviewing and editing this content, you’re just creating mediocre content faster. Worse, you could easily lose customers and audience because of AI’s trust issue. With the way many organizations today cut content design and strategists in their layoffs, many may have to learn the hard way that these experts are more critical than ever.