At this point, regardless of how marketers feel about it, AI is part of our toolkit. And like every tool, it’s important to know how to use AI properly. I’m saying this because as AI adoption grows, so do the AI gaffes and blunders.
It’s true that new tech always carries a learning curve, leaving plenty of mistakes in its wake. Since AI mimics content created by actual humans, however, its mistakes always seem a hundred times more egregious. When they’re made by your marketing team, the crew that’s meant to convey your message to your audiences, AI mistakes become even worse.
I’ve spoken before about the most common AI mistakes and how to avoid them. As AI continues to advance, though, it’s time to take a look at more AI gaffes you and your team need to avoid – and how to do it. Let’s take a look!
First things first about AI
The most important thing to remember about AI? You need to treat it like a very useful tool for your teams to use. Never treat AI like a replacement for your team itself.
Things AI can do: pull in facts and statistics (always check for links to the original sources), suggest topics and social media posts, and create generic social media replies.
Things AI cannot do: invent original content and, as we know all too well, AI cannot fact-check. In fact, AI has been known to make things up if it can’t find what the query asked for.
Now that we have that out of the way: here are three AI mistakes you need to avoid – and how to avoid them.
Failing to fact-check, fact-check, fact-check
AI has been known to be confidently wrong. Just one example? One AI-generated list claimed that doctors recommend pregnant women smoke cigarettes at least 2-3 times a day.
This is why everyone, and marketers especially, need to use AI with caution. It’s fine to have AI draw in a list of content suggestions. Since AI can’t determine fact from fiction, however, it’s still up to you to perform your due diligence and ensure the information you’ve received is correct.
This was driven home by a recent embarrassment in one of the US’s most storied newspapers. The Chicago Sun-Times published a summer reading list for 2025, full of thrillers, suspense and romance novels. There was just one problem: while all the authors were real, more than half the books were not.
When readers began posting that they couldn’t find the books for sale, the truth came out: a freelancer for King Features had used AI to generate the list of books. Instead of pulling in actual books, the AI generator simply created most of them instead – complete with cover images and plot synopses. And no one – from the freelancer to the King Features’ copy editors – fact checked the list before publishing it. Proof that at the moment, no one is immune to putting misguided trust in AI results!
Marketers should treat AI as a content aggregator, not the creator of the final product. You can absolutely use AI to find and create raw content for you. It’s still up to you to make sure the content it’s offering you is accurate – and actually exists.
Using AI to cut corners: you still need to measure twice and cut once!
The sheer volume of images that AI can generate has tempted many people to work far-far faster. Who didn’t dream of having a homework machine when they were a kid, after all? If you let AI do all the work for you and don’t bother to check the results, though, you’re going to end up with a living nightmare – for both you and your clients.
Again: you can have AI generate raw material for you, but you still need to edit it for accuracy, realism, and any mistakes.
- Check images very carefully. AI images can have a creepy uncanny valley feel to them. Yes, we’ve all heard the jokes about AI-rendered hands, but that’s just the beginning: pathways that look like they were designed by Escher, shadows falling in the wrong direction, and often, a shiny, unfocused veneer over everything. Turn to good old-fashioned photoshop to fix AI images, and if that doesn’t work, then consider not using them at all.
- Check the backgrounds of every image. AI generators have been known to pull in images of actual locations – but they might not be anywhere close to where you’re claiming they are. People are savvy; they’ll know if that beautiful beach café you claim is in Prince Edward County is actually on Lake Balaton in Hungary!
- Be aware of racist and sexist content. We’ve mentioned this before, but it needs to be said again: if offensive content is part of the AI algorithm, you could end up with an offensive image or social post. Don’t just look for blatant racial or sexist offenses either. Check AI content carefully for subconscious biases against marginalized groups – especially in AI images.
- It can be especially tempting to use an AI-generated social media calendar without vetting each post first. This can be an especially disastrous move. At best, you’ll have posts that sound robotic, and at worst, you’ll have a calendar riddled with false information and offensive claims about race, gender, culture or age. AI posts need to be copyedited in the same way as traditional ones.
Treat AI content the way a contractor treats a stack of wooden 2x4s: measure twice, cut once. Then you’ll have a final product that represents your message the way you wanted.
Making AI the pilot, instead of the navigator
Yes, AI can generate, in minutes, a volume of content that takes humans hours to create. This is why some companies have sometimes let AI replace actual people. Reducing your marketing team to data entry clerks (and no shade to data entry clerks) however, is a huge mistake.
We can’t say it enough: AI, for the moment, can’t understand your goals, strategies and tactics. Only human beings can do that, which is why you need to leave people in the driver seat of all your marketing campaigns. AI is very useful tool, but it should be treated like every other tool in your office. What AI produces still needs to pass quality control, and be managed by your team day in and day out.
- There is no “Set and Forget” mode in marketing. No matter how your content is produced, it still needs to be reviewed, edited, and monitored once it’s live.
- This is especially true for AI replies on social media. It’s fine to have AI-generated standard replies IF your team has reviewed and edited them.
- If you do choose to automate AI replies, you still need a human being monitoring the conversation on your socials. Be ready to step in once any interactions become too complex for AI to manage. YOU need to be aware if your brand is gaining traction!
- CONVERSELY: if your brand is blowing up for all the wrong reasons, then you need a human being to do some damage control ASAP.
Bottom line: you need to treat AI like an advisor, but never the captain of your mission.