Addressing the AI Elephant in the Room

Author: Jackson Ebeling

A practical look at where AI helps marketers and where human judgement still matters.

I've worked in digital marketing for close to ten years ow, which is a strange sentence to white when I'm still young enough to get carded buying cold medicine.

I built my first website at sixteen, optimized it for search, and ran ads for it. The site was hardly a technological marvel, but the timing gave me a front row seat to a lot of change. I watched advertising platforms become standard business tools. Website builders made development more accessible. Google grew powerful enough to shape how entire industries operated online.

So I remember digital marketing before generative AI was available to anyone with an internet connection. I'm also young enough to be buried in the technology now, testing where it helps and where it creates more trouble than it solves.

Clients and business owners keep asking the same kinds of questions. Does Performance Digital use AI? Will we let it create ads or branded assets? Is it going to replace designers, copywriters, strategists, agencies, or web developers? Should any of us be worried?

For me, those questions land in roughly the same place. How much trust should we give AI when it is representing a business in public?

Where I Draw The Line

Performance Digital uses AI. I use it every day. It helps sort information, organize rough thinking, support research, clean up writing, and handle repetitive internal work that would otherwise eat through hours.

The boundary comes when the work speaks for a client. We do not hand raw AI output to the public as finished creative, final copy, client communication, or strategy. A person who understands the business still has to make the choices, check the facts, and answer for the result.

This is a practical boundary. Ignoring AI would be fooling and, in my case, pretty hypocritical. I spend more time testing, breaking, and arguing with these tools than any functioning adult probably should. Giving them unchecked authority would be foolish too.

AI will change the way marketing gets done. It can make a capable person faster. It can also help a company produce an enormous amount of bland material before anyone stops to ask whether a customer will care.

Can AI make an ad? Of course it can. The harder questions begin after that. Should it speak for your brand? What will customers read into the choice? Who owns the mess if the answer goes badly?

Can AI Make a Good Ad

Almost every week, I hear some version of the claim that generative AI can replace a professional designer and produce ads, logos, website images, or other branded assets at the same quality or better.

I understand the appeal. Business owners are short on time, money, employees, patience, or some charming combination of all four. A tool that promises thirty seconds of work instead of a creative brief and several rounds of revisions is going to get attention.

But you know me, I like the inconvenient questions.

Look at what AI Companies Do For Themselves

OpenAI developed its first major ChatGPT brand campaign with its internal team, agency Isle of Any, and a full production crew. The team shot the campaign on 35mm film. ChatGPT helped with shot lists and schedules. People still handled the direction, cinematography, acting, production, and creative judgement. Anthropic's first Super Bowl campaign for Claude also appears in Mother London's agency portfolio.

These campaigns do not reveal every production decision, and the companies may have used AI elsewhere in the process. We can still see what happens when their reputations are on the line. They put experienced people around the work.

The companies selling these systems understand their strengths better than the rest of us. Their decision to keep human creative direction involved should tell us something.

Your Opinion is Not Customer Research

A business owner is usually not the audience for the ad. Something that looks acceptable from behind the desk may be the first contact thousands of people have with the company. They will make quick assumptions about its quality, credibility, personality, and value. Saying it looks good enough to me does not answer how a customer sees it.

A 2026 experiment by Ipsos and Syracuse University tested twenty ads with 3,000 consumers in the United States. Half were existing ads made by people. The other half were created entirely with AI from the same strategies. The human work performed 14 percent better for immediate effectiveness and 17 percent better for longer term effectiveness.

AI did best when the assignment was direct and functional. Here is the problem, here is the product, and here is how it helps. The gap grew when the ad needed emotion, humor, surprise, storytelling, or a recognizable point of view.

An AI system can assemble an advertisement. Giving somebody a reason to remember it is harder. Production was never the entire job. The job is to make a specific person care.

What Customers Read Into The Work

An ad made with AI does not have to look bizarre to hurt a brand. Sometimes the bigger risk is what the choice makes people assume about the company behind it.

The broadest recent data from the United States is uncomfortable. In an August 2026 survey from Bentley University and Gallup, 49 percent of Americans viewed business use of AI in advertising negatively. Only 19 percent viewed it positively. Negative views reached 66 percent among adults ages 18 to 29.

Consumers were more accepting when AI stayed behind the scenes and a person remained responsible for the result.

• 75 percent considered disclosed AI use acceptable for brainstorming or early drafts.
• 53 percent accepted disclosed AI use in final text, images, or video.
• 62 percent still considered synthetic people or voices unacceptable, even when the company disclosed their use.

People were much more comfortable with AI as a helper than as the face of the work.

Detection is The Wrong Test

People often defend AI creative by saying that customers cannot tell the difference. The research is mixed, which is exactly why detection makes a poor standard.

• A study of 1,276 people found that average accuracy across synthetic images, audio, video, and combined audio and video was 51.2 percent, about the same as guessing.
• A large public experiment gathered roughly 287,000 judgements. Participants correctly classified real images and images made with AI 62 percent of the time. That is better than chance, but far from reliable.
• In the Ipsos ad experiment, only 25 percent of people who saw an ad made with AI felt at least somewhat confident that AI was involved. Another 40 percent were unsure either way.

The same Ipsos viewers still rated the human ads as more imaginative, entertaining, unique, noticeable, and worth talking about. They reacted differently even when they could not explain why.

Sometimes the problem is easy to spot. A face, voice, movement, or expression slips into the uncanny valley. Other times the ad simply feels polished in a hollow way. It has no rough edges, no point of view, and nothing that could only have come from this business. Fixing the sixth finger does not fix that.

A customer may never consciously think that AI made the ad. They may scroll past, trust it a little less, or forget it immediately. Advertising does not get credit because the audience failed to identify the production method. It works when the audience notices the message and trusts it enough to respond.

When I see obvious AI output in place of thoughtful creative work, I do not think the company is more innovative. I wonder where else it is cutting corners. If this is the part the company chose to show me, what is happening the the parts I cannot see?

That assumption may be unfair. Customer perception has never needed to be fair to affect a business.

I call this brand cheapification. The company sees money saved in production. The audience sees a clue about how much care the company puts into everything else.

The Anger Does Not Stay With The Ad

Terrible advertising existed long before generative AI. I have seen plenty of campaigns made entirely by people that were confusing, irritating, badly timed, or apparently approved in a meeting where everyone was afraid to ask what the hell they were doing.

Some AI ads draw a different kind of criticism. People judge the decision behind the ad along with the ad itself. A complaint about the visuals quickly turns into a judgement that the company is lazy, greedy, dishonest, careless with its customers,, or eager to replace employees and creative professionals.

That gap is easy to see in IAB's 2026 survey of more than 500 Gen Z and Millennial consumers and 100 advertising executives. 82 percent of executives believed those consumers felt positively about ads made with AI. Only 45 percent of consumers actually did. Gen Z respondents were especially likely to describe brands using AI as inauthentic, disconnected, or unethical.

Another study in the Journal of Retailing and Consumer Services reviewed 7,822 comments on YouTube and Reddit about Coca Cola's holiday campaign made with AI. 81 percent expressed clear dislike. The discussion moved quickly from visual quality to soullessness, lost jobs, corporate greed, and judgements about Coca Cola itself.

Rival Technologies found a similar reaction in its July 2026 panel survey of 901 Gen Z consumers in the United States and Canada. 72 percent said they had taken at least one negative action after seeing marketing they believed was made with AI. Those actions included unfollowing, unsubscribing, complaining to other people, or buying less.

Those numbers need some boundaries. People who chose to comment on a controversial Coca Cola holiday campaign are not a representative sample, and nostalgia is practically a laboratory for making people angry. Rival asked panel members to report their own behavior. None of this proves that every ad made with AI will trigger a boycot.

It does help explain why the backlash can feel so personal. AI now carries arguments about jobs, authenticity, misinformation, environmental impact, creative work, and corporate cost cutting. A company may only mean to publish an ad. Some viewers will read that ad as the company's position on all of those arguments.

Gallup also found different levels of concern across political groups. Negative views of AI advertising reached 59 percent among Democrats, 49 percent among Independents, and 36 percent among Republicans. Majorities in all three groups opposed synthetic people or voices in advertising.

A bad ad can look like a mistake. An obviously automated one can look like a choice. Once the work is public, the company cannot control what people decide that choice says about it.

The Legal Questions Still Matter

Then we get to the considerably less entertaining subject of intellectual property.

Businesses sometimes talk about an AI logo, slogan, image, or campaign character as though possession of the file settles ownership. It does not. The vendor's terms, copyright protection, exclusivity, and legal clearance, each answer a different question.

The contract comes first. Terms differ by tool and by plan. OpenAI's current consumer terms say that the user owns the output as between the user and OpenAI, to the extent allowed by law.

Copyright is a separate issue. The United States Copyright Office says material created entirely by AI, or without enough human control over its expressive elements, is not protected. Human expression, a meaningful arrangement of material, or substantial modification may qualify depending on the facts. Propts along generally do not.

Ownership under a vendor contract also does not make the output exclusive. OpenAI warns that other users may receive similar material. The contract does not guarantee that an image, name, likeness, slogan, or design avoids someone else's right either. Important brand assets still deserve a similarity search and professional legal review.

None of this means every AI output is someone else's work wearing a fake mustache. A file that looks good is still not an originality opinion, a trademark search, or a guarantee.

The rist may be manageable for an internal mockup. A logo, mascot, slogan, or campaign asset that a business plans to own and defend for years deserves more care. That work needs meaningful human authorship and proper clearance.

I'm a marketer, not an intellectual property attorney, so none of this is legal advice. I do know the cheapest logo n the room becomes considerably more expensive after it goes on a building and a cease and desist letter arrives.

How Performance Digital Uses AI

AI has a job at Performance Digital. It also has limits.

What we use it for:

• Organizing documents, meeting notes, and internal information.
• Summarizing research and testing early ideas.
• Turning rough thoughts into a structure that a person can develop.
• Handling repetitive spreadsheet and internal data tasks.
• Helping edit writing that a person reviews and approves.

What still belongs to people

• Creating and approving finished advertising or public copy.
• Reviewing and sending client communication.
• Making final decisions about strategy, brands, or customers.
• Taking responsibility for the work.

AI can support the process around client work. It does not become the author or the final decision maker. Someone who knows the client and the audience develops the work, checks the facts, makes the call, and puts their name behind what goes out.

These rules may change as the technology, evidence, and law change. Any change will be deliberate and based on actual customer response. Discovering a shiny new button is not enough.

Yes, AI Helped to Create This Post

The disclosure is worth making plainly. If you couldn't tell from the clearly AI-generated thumbnail image, AI was partially used in the production of this post. It helped pull together research to support my observations from the real world, it helped organize my mess of notes on this topic into a cohesive message, and it also helped me challenge some of the internal claims and bias I had on this topic myself.

But the key is that I decided the primary message of this article, its supporting components, how it relates to our company, and then reviewed the outputs. Not AI. That is the kind of use being defended here.

AI will become a normal part of marketing work. Good professionals will use it to move faster and spend more time on the decisions that need them. It will also make it incredibly easy to produce cheap, generic material that nobody asked for and almost nobody will remember.

A weak idea stays weak when it is produced faster. Publishing the wrong thing still costs a business, even when making it was cheap.

AI has a desk in the office. The art department still keeps its own keys.