Now that the novelty of generative AI is wearing off, advertisers have gotten more serious about its usage — and are asking questions about its true effectiveness as they build a long-term strategy. As generative AI technology keeps maturing, we’re more able to gauge its impact across industries, refining its use and better understanding how to apply it.

For advertisers, “artificial intelligence” is often synonymous with “artificial creativity.” The goal, then, is to find the balance between generative AI creative work and human-generated creative work, which means gathering more information about how consumers perceive and respond to AI-made ads.

To that end, a new study conducted by researchers from Columbia, Harvard, Technical University of Munich (TUM), and Carnegie Mellon, using data from Realize (Taboola’s performance marketing platform), has found that AI and human ads perform remarkably well when run in tandem. Below, I’ll explain how best to use them together for maximum impact.

Why Performance Advertisers Should Balance AI Creative Work with Human Insight

By now, it’s clear that AI won’t replace human creativity and ingenuity. What it can do — and do well — is augment human time and effort, when used correctly. The study found that AI-created ads can even outperform human-created ads, which serves as a good reminder of AI’s incredible power and scale when harnessed intelligently.

The findings echo what many advertisers and marketers are finding in their day-to-day work: Generative AI can bring lots of benefits, but it needs human guidance to create the right outputs. For creative work, those outputs are ads that feel authentic and human — something that advertising and marketing teams can gauge.

Keep an eye on these trends as you’re using, measuring, and refining your own AI use in advertising.

1. AI-Generated Creatives Bring Efficiency

Efficiency is one of the top reasons why generative AI took off so quickly over the past few years, across industries. Advertising and marketing teams have embraced AI for creative work, even if they haven’t always fully understood how performance compares to human-generated work. For many of these teams, AI is the best option to help them keep up with competitors: 78% of respondents with ad budgets under $10 million are using AI tools for brainstorming and to develop initial basic concepts.

Cost efficiency is also now seen as a top benefit of AI, according to 64% of respondents in an Interactive Advertising Bureau (IAB) survey. In addition, 61% of respondents say that AI’s creative innovation is an advantage. AI-generated creatives can be produced at a fraction of the cost — estimated at just a few cents per request — and in significantly less time than human-generated work. Users are still exploring and understanding the nuances of using AI for core advertising and marketing work beyond efficiency.

2. AI vs. Human Ads Show On-Par Performance

Whatever the company size, team size, available resources, and current use of AI, advertisers and marketers have the same goal: improving performance metrics. What ultimately matters is how an AI-generated ad performs when compared to a human-generated ad.

In the controlled setting of the paper’s research, AI-generated ads maintained performance that was on par with human-made ads. The Columbia study found that ads using AI-generated images achieved “human-level” click-through rates (CTRs). And, if an AI-generated image doesn’t “look” like AI, the CTRs become “superhuman.” (Helpfully, the study also explored the boundaries of what’s perceived as artificial by ad viewers, such as overly stylized imagery.)

3. The AI-Human Relationship Is Still Evolving

AI use is extremely common in advertising work: 85% of those in the field are either using or planning to use gen AI to build video ad creative, with 85% using AI for social media ads, 73% for display, and 56% for TV ads. In the meantime, consumers have strong points of view on the use of gen AI ads: 60% think that the use of generative AI in ads should always be disclosed to consumers.

AI concerns aren’t unfounded, of course: One IAB survey found that 70% of marketers reported at least one AI incident, such as hallucinated outputs, biased or inappropriate content, or off-brand or offensive material. Because of those incidents, 40% of respondents had to pause or pull ads, while more than a third dealt with brand damage or PR issues.

Other studies have found that the disclosure of AI-generated content can negatively impact perceived credibility. These areas of exploration align with AI’s maturity curve, and offer plenty of opportunity for human advertisers to refine their work with generative AI in ads. According to one industry expert, the best-performing ad campaigns still rely heavily on human judgment, strategy, and experience.

4. Some Industries Are Adopting AI Tools Faster for Advertising

The Columbia study observed that AI adoption is a strategic choice, often led by performance-oriented advertisers. The AI image generation feature studied, Realize’s GenAI AdMaker, has been adopted early by high-performance industries, in particular:

• Pets: 15.6% adoption rate.
• Business: 4.68% adoption rate.
• Technology and Computing: 3.73% adoption rate.
• Real Estate: 3.52% adoption rate.
• Travel: 2.9% adoption rate.

In general, AI usage in advertising is set to grow, with 58% of respondents in one survey planning to increase the use of AI for creative generation in the next year. AI usage is expanding in terms of formats and workloads, too: Beyond developing concepts, AI can create net-new images and even video. Another IAB survey found that when it comes to ad creative, AI is mostly being used for scriptwriting (75%) and generating visuals (55%).

Key Takeaways

Generative AI usage in advertising has moved beyond initial experimentation into a phase of refining and strategizing, and a new study shows that AI-generated ads can match or exceed the performance of human-generated ads. For performance marketers, applying gen AI tools to ad creation allows them to scale massively beyond what a human team can do, by using data and previous learning to refine prompts and parameters to guide AI creation. Advertisers can build a parallel human-and-AI approach to refine their ad strategies and achieve new levels of performance.

Frequently Asked Questions (FAQs)

Does AI-generated imagery perform better in specific industries than others?

The Columbia study examined different industries and found major differences in which ones are adopting and benefiting from AI. Sectors like personal finance, food and drink, and pets all saw high adoption rates for using AI to create ads. Categories like education saw a much smaller adoption rate. This suggests that audience expectations and the typical look of an industry can influence how those audiences receive AI-generated content.

Can consumers actually tell when an ad is AI-generated?

The research found that nearly half (45%) of AI-generated ads were perceived as definitely or likely human-made, reflecting inaccuracies in consumer assumptions about AI creative. Raters identified 24.87% of actual human-made ad images as being AI-generated. This suggests that the perceived artificiality of an image — i.e., how much it looks like AI — is a more powerful driver of consumer behavior than the actual source of the image.

What are the specific visual tells that trigger Algorithm Aversion?

The study identified certain features in advertising that consumers tend to think are AI-created. Consumers generally associate high aesthetics, intense color saturation, and strong symmetry with AI-generated content. If an ad is overly polished, consumers may subconsciously or consciously sense its artificial origin and react negatively. But, AI-generated ads also disguise themselves by using features that consumers believe are human, such as larger facial areas and image clearness. Those features are actually more common in GenAI Ad Maker outputs than in human-made ones.