AI in advertising runs on your ideas. So, what are you feeding it?

Scroll through your socials for a few minutes. Count the ads that make you feel something. Now think about your ads. Do your posts just make people click, or do they make them feel? Do they just look or do they actually stop and read? Or do they ever wonder who made this and why? If they answered yes to any of these questions, you’re doing well.  

AI in advertising is everywhere now. More than half of marketers are already using generative AI for creative content and audience targeting. The tools are faster, cheaper, and more capable than anything that existed three years ago. But the output? Varies significantly. 

Now let’s ask ourselves, is the output good enough and if not, is it a technology problem or is it a thinking problem? Let’s have a look, because one of them is fixable right now.

What AI in advertising actually does 

AI in advertising uses machine learning, natural language processing, and data analytics to automate, create, and optimise marketing campaigns. It covers four areas:

Targeting and audience segmentation

AI analyses behavioural data, browsing habits, purchase history, content engagement, and identifies audience segments with more precision than any human analyst working manually. It finds patterns in data sets too large to process by hand and serves ads to people who are statistically more likely to convert.

Creative generation

Tools like Creatify and Arcads can turn a product URL or a short prompt into a full video ad in minutes. AI writes copy variations, generates image options, and produces hundreds of ad formats for A/B testing in the time it used to take to brief a designer and copywriter.

Programmatic bidding

AI runs auction-based ad buying in real time, adjusting bids, shifting budget between placements, and pulling spend from underperforming formats automatically. It operates 24 hours a day across platforms simultaneously.

Performance analysis and optimisation

AI monitors campaign performance continuously, adjusting creative elements, audience targeting, and spend allocation based on live data. It can predict which ad variations are likely to perform before a campaign launches.

These are real capabilities. The problem is what brands do with them.

Why most AI ads feel the same

AI learns from existing data. It generates outputs by identifying patterns across millions of examples and producing something statistically consistent with what has worked before. That makes it extraordinarily good at producing the average, middle-of-the-road execution that looks like everything else in the category.

Average is invisible.

The brief is where it goes wrong. When a marketer gives an AI tool a vague direction, "target millennial women interested in wellness, tone should be warm and aspirational", the tool does exactly what it's built to do. It produces the most statistically likely version of "warm, aspirational wellness." Which is what every other warm, aspirational wellness brand is already running.

Is AI really the problem. Or is it the marketer that handed it a brief with no real insight to it?

Strong advertising has always started with a specific human truth. Something uncomfortable or urgently true. That truth comes from real customer research, sharp strategic thinking, and someone willing to take a position the rest of the category won't. AI can’t manufacture that. It can only amplify what you give it.

If you give it nothing, it gives your audience nothing worth remembering.

The trust problem nobody is talking about 

A BCG study found 69% of consumers feel manipulated when brands use AI-generated content without disclosure. That’s most of your audience. The more AI-generated content looks slightly off, the more suspicion grows. 

The creative industry calls it the uncanny valley. AI visuals look almost real but not quite. Faces too symmetrical, environments too clean, movement slightly wrong. Audiences notice immediately, even if they can’t say why. It creates distance from the brand when you’re trying to connect. 

Several global brands faced backlash using AI campaigns without human oversight. The work felt made for an algorithm, not a person. 

Disclosure is coming. Regulatory discussions around AI transparency are accelerating in Europe and becoming relevant in South Africa too. Brands that act first will gain an advantage. 

What AI can’t replace 

AI won’t replace advertising. But it will replace specific tasks within advertising. 

It can’t generate a genuine strategic insight. It can’t sit across from your customer and notice the thing they didn't say. It can’t read a one-star review and extract the emotional subtext. It can’t take the uncomfortable strategic position that makes your brand different when the rest of the room wants to play it safe. 

It also can’t take accountability. When a campaign fails, someone must own it, learn from it, and fix the next one. AI optimises toward the metrics it's given. If those metrics are wrong, or if the brief was wrong before the metrics were set, no algorithm will tell you. 

Human judgment sets the ceiling. AI executes within it. 

You’d think the brands winning with AI have the best tools, they don’t. They did the strategy first, then used AI to say it louder, faster, further than they could alone.

How to use AI without losing your edge 

Strategy first. Define the one true thing your brand can say that your competitors can't claim. This requires human thinking, actual research, customer conversations, category analysis, and the strategic courage to take a position. Not something you outsource to a tool. 

Concept second. Develop the creative territory before anything gets generated. What's the tone? Where's the tension? What will make a real person stop and think? This is where human creativity sets the ceiling on everything that follows. 

AI third. Once you have a clear brief built on real insight, use AI to execute at pace. Generate variants, test copy angles, produce visual options, run A/B tests across segments. High-volume, high-speed execution within a defined creative direction, that's where AI earns its place. 

Human judgment throughout. Every output gets a human eye before it reaches an audience. Not to check for typos, to check for truth. Does this feel real? Does it say something a real person would care about? 

Build a feedback loop. AI optimises for the data it receives. Track the right metrics. With high clicks but low brand recall, you train AI to optimise for the wrong outcome. 

The South African opportunity brands keep missing 

Global AI tools use global data. They produce global-feeling work. This is fine for multinationals. It is a liability for brands in South Africa. Context matters. 

South African audiences are diverse. Load shedding affects when people are online and what patience they have with video ads. Language switches in conversations are normal. Price sensitivity varies sharply across provinces, income, and age. 

What works in Sandton may fail in Soweto. Cape Town may miss Polokwane. Remember, cultural reference points, humour, trust signals are local. 

A brand that does research first, then uses AI, produces work that feels made for the right person. That specificity is human work like the work we do at DC Lab. 

The question to ask before your next campaign 

Before you open any AI tool, answer this in one sentence: what do we believe that our competitors don't have the courage to say? 

If you can't answer it, the brief isn't ready. Go back. Talk to real customers. Find the uncomfortable truth. Take the position. 

Then use AI to execute it faster and further than you ever could before. 

DC Lab works with brands that want to move fast without losing their edge across paid, SEO, content, and direct marketing. If your current marketing feels like it could belong to anyone, let's change that.