Quick Overview
Do AI-generated ads actually work? AI-generated ads can work, but performance varies based on the format, quality, disclosure requirements, and level of human oversight involved. Current research suggests AI is more reliable as a tool for experienced advertisers than as a replacement for them.
Considering using artificial intelligence (AI) to create your own online ads or have concerns about your pay-per-click (PPC) agency using AI to generate ads? It’s smart to pause. While research is still emerging, it’s generally not a good idea to pass everything off to bots just yet, but there are many ways AI can help you streamline the process or improve performance. We’ll take a look at where it excels (and where it doesn’t) below.
AI-Generated Ads Can Negatively Impact Customer Perception and Trust
Nearly half of all Americans expressed negative sentiments about the use of AI in ads in a recent Gallup poll. The study found that just 19 percent express positive sentiments, while 49 percent respond negatively. While the data doesn’t transfer globally, the sentiment is largely shared across most Western nations, YouGov data shows.
Opinions vary based on how AI is used. For instance, 51 percent said they’re uncomfortable with brands creating virtual ambassadors. Using AI to edit product images or design product images is also a no-go, with 48 percent and 47 percent of the population reporting discomfort. Other areas, such as copywriting and deciding where to place advertisements, produced a more even split.
More strikingly, 78 percent say AI makes ads feel less authentic, American Association of Advertising Agencies (4As) research shows. The same study also found that 63 percent are less likely to purchase from a brand that uses AI-generated ads and 73 percent are less likely to trust an ad if they suspect it was created with AI.
Sales and Performance Results Are Mixed with AI-Generated Ads
Research on AI-generated ads shows some promise, but the data is often spotty and, even with success stories, still shows a strong need for human involvement.
AI-Generated Ad Images Sometimes Perform Well
For instance, most major studies focus on generative artificial intelligence (GenAI) ad design, not the ad copy. This includes one from New York University (NYU) that rotated three groups of images in a Google campaign: human-created designs, GenAI-modified designs, and GenAI-created designs. In this example, the GenAI-created designs outperformed both the human-created and GenAI-modified designs in terms of clicks, NYU researchers say. However, these were very basic graphics for shopping ads featuring little more than haircare products in a tropical setting. Moreover, the AI tools weren't given the same restrictions and briefs as the human creatives, and the top-performing ads also included redesigns of the products themselves. It’s impossible to say which would have won if the playing field had been level.
Another study was run by Columbia, Harvard, TUM, and Carnegie Mellon. In this example, AI-generated images had a raw click-through rate of 0.76 percent, compared to 0.65 percent for human-created images, researchers say. However, after accounting for differences within campaigns, AI and human images performed comparably overall. AI images performed best when viewers believed they were human-created
Limited Tests Favor Human-Crafted Ad Copy
Research comparing AI-generated and human-crafted ad copy is surprisingly limited. However, one direct Google Ads test found that human-crafted copy outperformed AI-generated copy by a wide margin, as Search Engine Journal reports.
| Metric | AI-Generated Ad Copy | Human-Generated Ad Copy |
|---|---|---|
| Impressions | 713 | 1,306 |
| Clicks | 26 | 65 |
| Click-Through Rate (CTR) | 3.65% | 4.98% |
| Average Cost per Click (CPC) | $6.05 | $4.85 |
Additionally, there’s an older study by Microsoft. Their research found that AI working alone produced ad copy that received 28 percent fewer clicks. However, when AI was brought in to rewrite human-generated copy, the ads received 32 percent more clicks.
A similar pattern appears in research from Meta. Researchers compared two AI models built on the same foundation. One was trained to imitate curated examples, while the other received additional training based on historical click data. The performance-trained model increased click-through rates by 6.7 percent, Meta researchers say. The findings don’t compare AI with human-crafted copy, but they demonstrate how the information AI receives can influence ad performance.
AI-Generated Video Ads Are Most Likely to Backfire
When people were shown AI-generated video ads as a test, they identified most of them as being AI-generated without being asked, Nielsen reports. More importantly, they also rated them as being more annoying, boring, and confusing. Additionally, researchers found that the ads were less memorable, particularly as quality declined.
Similar findings are reported by Harvard Business Review (HBR). However, their study indicated that AI videos underperformed even when people didn’t realize they were AI-generated, noting that they had 14 percent less short-term sales potential and were 17 percent weaker on long-term brand equity. They point to a pair of Chewy video ads as proof. Although both follow the same playbook and are of good quality, the AI-generated video is more generic and less memorable.
Human-Created
AI-Created
AI Disclosure Can Kill Advertising Results But is Often Required
One recurring theme in the studies was that, when people believed what they were viewing was AI-generated, they liked it less and responded to it less. AI-generated ad creative performs best when it’s very high quality or indistinguishable from human-generated creative. However, brands are increasingly being required to disclose the use of AI.
For instance, the EU AI Act requires disclosure of AI-generated or manipulated content that qualifies as a deepfake. This primarily covers realistic images, audio, and video that could be mistaken for authentic people, objects, places, or events. Meanwhile, New York State has a law that states it must be disclosed when AI-generated synthetic performers are used. India and China take it a step further and not only require AI disclosures, but also the inclusion of traceable metadata.
For advertisers, disclosures like these can reduce effectiveness by as much as 31.5 percent, per NYU. In other words, even if you do have top-notch AI-generated ad creative that people cannot easily identify as such, you may still be required to disclose that it’s AI-generated, which can ultimately harm the results.
AI Works Better as a Tool to Help You Scale and Target
Another point of crucial importance is that all these studies typically involve experts in the process. When generative artificial intelligence (GenAI) is being used to create graphics, for instance, it’s still professional designers using the tool and selecting which designs to use in the ads. Similarly, when ad copy is being tested, professional copywriters are typically guiding the tools and choosing which variants to publish. They’re not pulling in busy business owners or their sales and ops teams to craft AI-generated ads. They’re bringing in seasoned pros with decades of experience.
With that in mind, AI cannot replace a PPC agency or your in-house team. It can, however, help with things like ideation and iteration, so you can scale and test faster. It can also help you improve targeting and personalize your ads.
Get Better Results Faster with a PPC Agency
AI-generated ads can work, but they need to be created by experts, managed by experts, and overseen by someone who understands the disclosure laws where you operate and advertise. That usually means having a full team working on your advertising. An experienced PPC agency can help you identify the best path forward and get the results you deserve regardless of whether AI-generated ads fit your business needs. If you’d like to be matched with a suitable agency, connect with us for a complimentary matching consultation.
FAQs on Using AI-Generated Ads
What’s the Difference Between AI-Generated and AI-Assisted Ads?
AI-generated ads contain copy, images, audio, or video substantially created by AI. AI-assisted ads are primarily created by people, with AI used for tasks such as brainstorming, editing, resizing, or producing variations. This is an important distinction, as some platforms and jurisdictions require disclosures for substantially AI-generated or manipulated content.
Can AI Create and Manage an Entire PPC Campaign?
AI can generate ads, recommend audiences, adjust bids, distribute budgets, and identify performance trends. However, it doesn’t understand your customers, margins, sales process, or brand as well as an experienced team does. It also can’t accept responsibility for false claims or wasted spend. A qualified person should supervise every campaign.
Are AI-Generated Ads Cheaper to Produce?
They can reduce the time and cost required to produce initial concepts and variations. However, those savings can disappear if your team must correct inaccurate copy, repair strange visuals, or test large volumes of weak creative. It’s also important to weigh what happens after production. Metrics like clicks and sales influence the value of what you’ve produced. Your ads can have an impact on things like brand affinity and loyalty as well. When these things are taken into account, AI-generated ads may not bring the same value as human-created ads even if they are cheaper to produce.
Are AI-Generated Ads Allowed on Google, Facebook, and TikTok?
Generally, yes, but the usual rules governing misleading claims, intellectual property, restricted products, and impersonation still apply. TikTok requires disclosure for significantly AI-generated or altered ads, Meta labels some ads created with its AI tools, and Google requires disclosures for certain synthetic content in election advertising.
Can I Use an AI-Generated Person or Spokesperson in an Ad?
Yes, but don’t present a synthetic person as a real customer, expert, or employee. Avoid copying a real person’s appearance or voice without permission, and disclose synthetic performers where required. Any testimonial must still be truthful and reflect a genuine experience.
Who Is Responsible if an AI-Generated Ad Is Misleading?
Your business is still responsible for the claims made in its ads. Blaming the AI tool, advertising platform, or agency won’t make an inaccurate claim acceptable. Agencies and other parties may also face liability depending on their involvement, but advertisers should verify every factual claim and retain supporting evidence before publication.
Can AI-Generated Ads Be Copyrighted?
In the United States, material created entirely by AI generally isn’t protected by copyright. Human-created elements, creative arrangements, and meaningful modifications may qualify. Your prompts probably aren't enough on their own. Because protection depends on the human contribution, keep records showing how your team selected, edited, arranged, and developed the finished ad.
How Much Human Review Do AI-Generated Ads Need?
Every ad should be reviewed before publication. Someone needs to verify claims, prices, product details, images, trademarks, targeting, landing-page consistency, and disclosure requirements. High-risk industries such as healthcare, finance, employment, and legal services require additional scrutiny. Review should continue after launch because strong click data doesn’t guarantee qualified leads or profitable sales.
What Should I Ask My PPC Agency About its Use of AI?
Ask which tools it uses, what campaign tasks they perform, and whether your business data is entered into third-party systems. You should also ask who reviews the output, how mistakes are caught, how disclosure requirements are handled, and whether AI-generated ads are measured separately from human-created ads. The agency should be able to answer clearly.
Which Parts of PPC Advertising Are Best Suited to AI?
AI is particularly useful for brainstorming, creating ad variations, resizing creative, analyzing search terms, identifying patterns, summarizing results, and supporting bidding or targeting decisions. It’s less suited to developing your core strategy, defining your offer, making sensitive brand decisions, or approving factual claims. Those areas require business context and human judgment.
How Can I Test AI-Generated Ads Against Human-Created Ads?
Keep the audience, offer, landing page, budget, placement, and campaign objective consistent. Change only the creative you’re testing, and give each version enough traffic to produce meaningful results. Compare conversions, qualified leads, cost per acquisition, and return on ad spend. A higher click-through rate doesn’t automatically mean the ad generated better business.
