The rapid adoption of visual generative AI (genAI) in marketing and advertising is undeniable. By 2024, nearly 90% of marketers reported using genAI tools, with 45% specifically adopting them for visual tasks like social media and website content. This enthusiasm, however, has outpaced a critical, underexplored question: While genAI promises efficiency and scale, are the ads it produces actually effective? A recent comprehensive study offers startling answers that run directly counter to the best practices established for text-based AI assistants. The findings reveal a clear, evidence-based playbook for using visual AI and show that the most common approach is a strategic dead end. AI Excels at Creating from Scratch, Not Editing an Expert's Work The study's central and most surprising finding is that advertisements created entirely by visual genAI from a text prompt significantly outperformed ads created by human experts. In a field study, these "genAI-created ads" increased click-through rates (CTR) by up to 19% compared to the human-created benchmarks. This is the study's most critical insight for practitioners. The mental model we've built from using text-based AI like ChatGPT (where AI is an excellent editor) is not just unhelpful for visual AI; it is precisely the wrong approach. While textual AI models excel at refining existing content, this research shows the opposite is true for visual genAI. The study notes this fundamental difference: In contrast, our findings reveal an opposite pattern for visual genAI: rather than benefiting from task decomposition (e.g., editing or modifying), visual genAI is more effective when producing complete outputs as unified creations. 2. Trying to "Enhance" a Human-Made Ad with AI Is a Trap The study also investigated "genAI-modified ads," where AI was used to enhance or edit advertisements originally designed by experts. The results were stark: these AI-modified ads showed no significant improvement in the field study and, in a controlled lab setting, were even found to reduce consumer purchase intentions. The researchers discovered that when genAI is constrained by an existing image, it struggles to maintain a sense of real-world authenticity. These genAI-modified ads "fail to preserve ecological validity," which ultimately harms their effectiveness. Creative Freedom Is the Key to Success The research revealed a direct relationship between the creative freedom given to the AI (what the study calls "output constraints") and the performance of the final ad. The less constrained the AI, the better the results. The study uncovered a clear performance hierarchy: High Constraint: Modifying an existing ad. This performed the worst.
Low Constraint: Creating a new ad from scratch based on a text prompt. This performed significantly better, boosting CTR by up to 19%.
Lowest Constraint: Creating a new ad and also designing the product packaging within it. This amplified advertising effectiveness even further. High Constraint: Modifying an existing ad. This performed the worst. High Constraint: Low Constraint: Creating a new ad from scratch based on a text prompt. This performed significantly better, boosting CTR by up to 19%. Low Constraint: Lowest Constraint: Creating a new ad and also designing the product packaging within it. This amplified advertising effectiveness even further. Lowest Constraint: This hierarchy provides a clear directive: to maximize ROI on visual AI tools, treat them as unconstrained ideation engines at the beginning of the creative process, not as constrained finishers at the end. The High Cost of AI Transparency The study delivered a final, critical finding regarding AI disclosure: telling consumers that genAI was involved in an ad's creation significantly reduces its effectiveness. The impact was dramatic. In the field study, AI disclosure led to a 31.5% decrease in CTR relative to human expert ads. This creates a critical trade-off for businesses and policymakers. With the growing push for AI transparency regulations, such as the EU's AI Act which mandates disclosure for AI-manipulated content, this research quantifies the potential business cost of complying with such mandates. 31.5% decrease in CTR The research provides a clear, evidence-based roadmap for integrating visual genAI into the advertising workflow. The most effective approach is not to treat these tools as simple editors for human work, but as powerful, unconstrained partners for creative ideation. By giving genAI the freedom to create holistic concepts from the ground up, marketers can unlock significant performance gains that surpass even expert-level human creation. This leaves us with a final, thought-provoking question for the future: As AI-generated content becomes ubiquitous, will this "disclosure penalty" fade away, or will audiences always place a higher value on perceived human creativity? Podcast: Podcast: Podcast: Apple Podcast: HERE
Spotify: HERE Apple Podcast: HERE Apple Podcast: HERE HERE Spotify: HERE Spotify: HERE HERE

The is an opinion piece based on the author’s POV and does not necessarily reflect the views of HackerNoon.

Surprising Study Reveals the Right Way to Use Visual AI in Advertising

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Untitled Story

3 Experiments That Reveal the Shocking Inner Life of AI Introduction: Is Anybody Home?

The Noonification: How Often Do NFTs Pass The Howey Test? (1/13/2023)

Darwin's Hybrid Intelligence to Align AI & Human Goals for Startups & VCs

The Noonification: White Man (11/26/2022)

The Noonification: The Metaverse is a Sh*tshow (11/2/2022)

3 Experiments That Reveal the Shocking Inner Life of AI Introduction: Is Anybody Home?

The Noonification: How Often Do NFTs Pass The Howey Test? (1/13/2023)

Darwin's Hybrid Intelligence to Align AI & Human Goals for Startups & VCs

The Noonification: White Man (11/26/2022)

The Noonification: The Metaverse is a Sh*tshow (11/2/2022)

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps