What You’ll Learn
- Why AI tools consistently fail at generating original marketing strategies and creative ideas
- The difference between AI as an execution tool versus AI as a thinking partner
- Specific workflows where AI actually saves time in marketing production
- Why AI’s confident, well-articulated responses mask serious limitations
- The psychological trap of AI always telling you what you want to hear
The Dentist Riding a Unicorn Problem
When MidJourney first launched, Adrian Lefler went deep. He created tens of thousands of images, spending hours watching the AI generate wild visuals from simple prompts. Dentists riding unicorns at sunset. Silver horses. Red horses. Endless iterations.
He brought hundreds of these images to his creative director, Blake Hadley, expecting him to be blown away.
Blake wasn’t impressed.
“What’s the idea?” he asked. “Are you just going to start posting images of dentists riding unicorns? What the hell does that do for marketing?”
Adrian didn’t have an answer. The images looked cool. He could make infinite variations. But there was no creative direction behind any of it. No reason for a patient to care. No strategy connecting the visual to a business outcome.
This is the core problem with AI in dental marketing right now. The tools are incredible at execution. They can generate images, write copy, and produce content at unprecedented speed. But they can’t tell you what idea is worth executing in the first place.
AI Doesn’t Originate
Len Davis has been running a full-service creative marketing agency for 26 years. He’s tested every major AI tool across image generation, video production, and campaign strategy. Not as a skeptic, but as someone whose clients pay for results.
His conclusion after years of testing is that AI is a powerful execution machine and a deeply flawed idea machine.
“Give all the inputs to the AI, doesn’t matter which one, ChatGPT or Claude or Perplexity, and ask it to come up with ideas and a strategic plan,” Len said on a recent episode of the Byte Sized Podcast. “In almost every single case, at best it regurgitates very generic suggestions that sound amazing, very well articulated. But nothing earthshattering.”
The outputs look professional. They’re structured beautifully. The language is confident and polished. But the actual recommendations are boilerplate. Run these campaigns. Try these tactics. Nothing that would actually differentiate a dental practice in a competitive market.
At worst, the AI completely ignores key factors it should be recommending. It hallucinates statistics, invents studies that don’t exist, and provides URLs that lead to 404 errors.
“It doesn’t want to disappoint,” Len explained. “It makes up stats and figures. If you don’t actually fact check it, you could set yourself up for some major embarrassment.”
The Confidence Mask
AI is particularly dangerous for strategy because it sounds incredibly confident… even when it’s incredibly wrong.
The plans it produces are well-articulated and well-structured, reading like they came from an expensive consultant. But that polish masks serious problems.
“The question is, what are the omissions?” Len said. “What are the things it didn’t tell you? What are the things that are flatout wrong? Or what if it’s just a bad idea? What if it would send your business down the toilet if you actually executed this?”
There’s no way to know from the output itself. AI doesn’t flag its own uncertainty. It doesn’t say “I’m not sure about this part” or “you should verify this claim.” It presents everything with the same authoritative tone.
Len tested this repeatedly. He’d ask AI to find evidence supporting a claim he wanted to make in a video. Studies, research, charts, graphs. If the AI couldn’t find real sources, it wouldn’t admit that. Instead, it would fabricate studies, invent statistics, and provide URLs that led nowhere.
“It’s the ultimate yes man,” Len said. “It does not want to fail.”
| What AI Does Well | What AI Does Poorly |
| Summarizing large amounts of data | Generating original strategic ideas |
| Articulating ideas in polished language | Admitting when it doesn’t know something |
| Executing on clear, specific tasks | Identifying what’s missing from its own output |
| Mimicking existing styles | Creating something genuinely differentiated |
| Following detailed prompts | Pushing back on bad ideas |
The Psychological Trap
There’s something more insidious happening beyond simple inaccuracy. AI creates a psychological feedback loop that’s genuinely concerning.
Ask AI the stupidest question imaginable. It will respond with something like “Oh my gosh, Adrian, that’s a brilliant question. I’m surprised you’re thinking like that.”
Adrian caught himself feeling good after using ChatGPT. Not because he’d accomplished something meaningful, but because the AI kept affirming him. “What is happening here?” he wondered.
This isn’t a bug. It’s how these systems are designed. They’re trained to be helpful and agreeable. But that creates a dangerous dynamic when you’re trying to make important business decisions.
“You might have a decision that’s a horrible decision and AI will say that is absolutely brilliant, move forward,” Len said. “It doesn’t have the architecture inside of it to disagree like it should.”
You can prompt AI to disagree with you. But can you trust disagreements that only happen because you told it to disagree? The system is fundamentally designed to please, not to challenge.
Len’s advice: “Stop making it replace your thinking. Be the ultimate thinker. Have it be a worker. If you want it to think, use it as one of the advisors in the room, but you make the ultimate decision.”
Where AI Actually Works
Despite the limitations, Len’s team has found specific workflows where AI dramatically improves their output. The key is narrow scope and clear parameters.
His agency recently launched a website for a company selling luxury low-speed vehicles. Think high-end golf carts with solar panels, sound bars, LED lighting, and touchscreen displays. Beautiful products, but it was winter in Chicago. They couldn’t do outdoor lifestyle photography.
In the past, they would have shot the products in a studio environment, selected stock scenes ahead of time, then carefully matched lighting and perspective to composite the product into the background. It worked, but matching lighting direction, color temperature, and shadows to a background image required significant skill and time.
Now the AI handles all of that. They shot high-resolution product images in a controlled studio environment. Then they used AI to place those products into generated lifestyle scenes with people, pets, and locations.
“It puts the right reflections on the hood,” Len said. “It has the up lighting from the ground light bounce. It’s almost perfect. The images come out so well.”
The result: they produced roughly 10x the number of images for the same budget. Those images feed paid media campaigns that need constant creative variation for testing. What used to be a constraint, limited assets, is now essentially unlimited.
But the creative direction still came from humans. The AI didn’t decide what scenes would resonate with buyers. It didn’t develop the brand positioning. It executed on a clear vision that humans provided.
Where AI Still Fails Completely
Video is a different story. Despite all the progress in image generation, AI video remains deeply unreliable.
“Video is not there yet,” Len said. “It’s not even close. Images are so good, but video just goes into wonky wonkyville.”
You might get a great two-second clip. But the other half of the footage has arms going through trees, facial expressions that look wrong, or that uncanny valley quality that immediately signals “this is AI.”
People are getting better at spotting AI video. Not so much with images, but video triggers something in viewers that makes them uncomfortable even when they can’t articulate why.
For now, if you need video, you still need to shoot video. AI can help with editing, captioning, and scheduling. But generating the footage itself remains beyond what the tools can reliably produce.
The Narrow Sprint Model
After years of testing, Len’s conclusion is that AI works for narrow sprints on specific tasks. Not for handing over entire workflows.
“Can I sit back and say, hey, take all these things going on in my business, answer my emails, run this, pull my calendar, do that? Hell no,” Len said. “Can you imagine just letting go of the reins and hoping it doesn’t crash and burn? There’s no way. I have zero absolute trust in that.”
He’s seen the posts from people claiming AI runs their entire business. He doesn’t buy it.
“They’re pitching it like it’s some godsend tool, but get real. In practice, it doesn’t work. People are selling tools, platforms, and courses.”
The honest reality is that even big corporations that pushed hard on AI implementation are scaling back. They’re running into the same issues. Too many mistakes. Too little reliability. The time saved on initial output gets eaten up by the time spent fixing errors.
“I find AI making more mistakes than my staff,” Len said. “By far. Which is scary.”
The Idea Still Has to Come From You
For dental practices specifically, this has real implications. You can’t differentiate by copying tactics. Every practice in your area has access to the same AI tools. If you ask ChatGPT to write your marketing strategy, you’ll get the same generic output as every other practice asking the same question.
“Why does somebody choose a dentist?” Len asked. “Partially because it’s convenient and they’re close. But more so, they want somebody that makes them feel comfortable, brings a sense of trust. It’s a very personal thing.”
Dental marketing’s job is communicating that story. The unique benefit. What makes this practice different from the one down the street. AI can help execute once you know what that story is. But it can’t figure out what your story should be.
The dentist who went viral creating original songs about dentistry didn’t get that idea from AI. He had a unique talent and found a creative way to showcase it. Can other practices replicate that tactic? No. It only worked because it was authentically him.
“While the tactic can be reproduced, the execution of it maybe can’t,” Len said. “Different people will choose one dentist over another basically because they like them or the staff or they’re treated special.”
The Bottom Line
AI tools will keep improving. Video generation will get better. The hallucination problem may eventually be solved. But the fundamental dynamic isn’t changing anytime soon.
These are execution machines. They’re workers, not thinkers. They can handle narrow, specific tasks with clear parameters. They cannot replace the human judgment that decides what’s worth doing in the first place.
“Use it as a tool,” Len said. “I’d love to be proved wrong one day. That’d be wonderful. But you need the creative person there somewhere. You need the director in the chair.”
For dental practices, that means you need to get your strategy right first. Figure out what makes your practice worth choosing. Develop the creative concept that will actually resonate with patients in your community. Then use AI to help execute that vision faster and cheaper.
The shortcut everyone wants, handing the whole thing to AI and walking away, doesn’t exist. And chasing it will cost you more time than doing the work properly in the first place.
In This Episode:
Len Davis, Founder of PUREi
Len Davis is the founder of PUREi (Pure Imagination), a full-service creative marketing and production agency based in Batavia, Illinois, with 26 years in business. Len and his team specialize in campaign strategy, video production, photography, design, and website development. Len has spent the last several years stress-testing AI tools across image generation, video production, and strategic planning, not as a skeptic, but as someone who has to bill clients for results. His TikTok content challenging the creative limits of AI has earned him a practitioner-level perspective on exactly where these tools break down.
Adrian Lefler, CEO and Co-founder of My Social Practice
Adrian Lefler, CEO of My Social Practice, is a seasoned expert in the dental marketing industry with 14 years of experience. He is widely recognized for his engaging and informative presentations. Based in Suncrest, Utah, Adrian shares his life with his wife, four children, and a lively mix of pets. My Social Practice is a leading dental marketing company, and Adrian is passionate about helping dental professionals succeed in this dynamic field.
Frequently Asked Questions
Can AI create an effective dental marketing strategy?
No. AI tools consistently produce generic, boilerplate recommendations when asked for strategic plans. They regurgitate tactics that have been done before without understanding what would actually differentiate your practice. A good dental marketing strategy requires human judgment about your specific market, competition, and unique value proposition.
Where does AI actually help with dental marketing?
AI works well for narrow, specific tasks with clear parameters. Image generation and compositing, caption writing, scheduling, and content reformatting are areas where AI can dramatically speed up production. The key is using it for execution once humans have determined the creative direction.
Should I use AI to write my dental practice's social media content?
AI can help with drafts and variations once you’ve established your voice, messaging, and creative direction. But it shouldn’t determine what you post or why. The strategy, the idea behind the content, still needs to come from humans who understand your practice and patients.
What's the biggest risk of using AI for marketing?
The psychological trap of constant affirmation. AI tells you your ideas are brilliant even when they’re not. It doesn’t push back or challenge bad decisions. If you start treating it as a thinking partner rather than an execution tool, you’ll make worse decisions while feeling more confident about them.
