AI has become one of those tools that I reach for fairly regularly, although mostly as a research tool. It's almost like a more conversational search engine, or what I use when I’m working through ideas for visual designs. I’ll use it to find information, explore different directions or help me think through something I’m creating. But even then, once the answer looks useful, there is always that temptation to go, "Cool, thanks,” and move on.
Then, at the Modern Marketing Expo, someone asked a question I somehow hadn’t thought to ask before: “Have you ever asked your AI to review its own work?”
The expo had already been messing with my idea of what marketing looked like. In my head, Modern Marketing Expo meant digital marketing. I was expecting AI tools, Meta Ads, customer relationship platforms and probably enough software demos to make my laptop nervous. Instead, I walked into banner printers, embroidery machines, 3D printers, different materials and machines dedicated to parts of marketing I had never really stopped to think about before. At one stall, I spoke to someone who builds racing simulators, and he casually explained that several parts were 3D-printed. I’m sorry, what?
Somewhere between the embroidery machines and the racing simulator, I realised my idea of what marketing looked like was much smaller than the actual industry. I had been thinking screens, ads and campaigns. Meanwhile, there was this entire physical world of branding and production happening right in front of me.
Then came the question.
Later in the day, I attended a talk by Tendai Rukwava, founder and CEO of Coral Communications, a Johannesburg-based PR, marketing and communications agency. According to IPREX’s profile of Coral Communications, the agency has worked on strategy-led campaigns across Africa, the UK and the US and was recognised as one of South Africa’s Top 10 PR agencies in a 2026 independent poll. Rukwava was also named Leading Communications Expert of the Year at the 2024 Influential Business Woman Awards. So, it’s safe to say, when she started talking about responsible AI use in marketing, I was listening.
Out of everything she spoke about, though, one question followed me home: Have you ever asked your AI to review its own work? I hadn’t. Even though I use AI regularly for research and visual ideas, I realised I had a habit of treating the first useful-looking answer as the correct answer. I had never really stopped and asked the tool itself: What’s weak here? What did you miss? What assumptions did you make? Is this actually the best work you could have produced?
So I tried it. I took something AI had already helped me produce and asked it to critically analyse its own answer. Suddenly, the same tool that had confidently handed me the first version started pointing out generic wording, missing context and places where the thinking could have gone deeper. The interesting part was that I didn’t agree with everything it criticised. Some suggestions were useful. Others would have taken the work in a direction I didn’t want. That was probably the bigger lesson: asking AI to critique itself doesn’t make it the final judge. It just gives you another perspective to work with.
Maybe the skill is knowing when to push back. I’m still at a stage in my career where I’m constantly looking for more: another course, another tool, another framework, another person who knows something I don’t. The expo definitely gave me plenty of that, but what I didn’t expect was to leave with a better way of questioning something I already use almost every day.
AI can still miss things, make assumptions and sound extremely confident while doing it. Asking it to critique itself doesn’t magically fix that. You still have to decide what makes sense, what doesn’t and what is actually useful for the thing you’re trying to create. Maybe using AI well is partly about knowing when not to accept the first answer. Sometimes you just need a better question.
I went into the Modern Marketing Expo expecting to learn about new tools, and I definitely did. It’s difficult not to when someone casually tells you they used a 3D printer to help build a racing simulator. But the thing I’ve actually used most since leaving wasn’t another piece of software or some complicated marketing technique. It was one question I could take home and use immediately.
Sometimes the most useful thing you learn is just a better question.