There was a time when doctors had WebMD to worry about.
A patient would walk into the office convinced that their headache, fatigue and sore shoulder were clear evidence of a rare neurological disorder they discovered during three hours of internet research.
Doctors hated it.
Now imagine that same patient walking into the office with a professionally written differential diagnosis, a recommended list of tests and treatment options, and a series of questions challenging the doctor’s conclusions.
Welcome to the age of AI.
And doctors aren’t the only professionals dealing with it.
The Distance Between Curiosity and Confidence Has Collapsed
Artificial intelligence has given us access to an extraordinary amount of information. I use it. Our agency uses it. I highly encourage businesses to use it.
But something else is happening that more people should acknowledge.
AI has dramatically shortened the distance between knowing very little about a subject and feeling extremely confident about it.
In the past, developing a strong opinion about something outside your expertise required some work. You had to read, research, talk to people, and maybe even hire an expert. You had to spend enough time with the subject matter to discover that the answers to your questions were more complicated than you thought.
That friction of time and effort resulted in a better understanding of the nuances around a topic.
Today, you can ask an AI platform a question and receive a polished, organized and remarkably convincing answer in seconds.
It has headings. It has bullet points. It explains its reasoning. It may include statistics and citations. It sounds like it knows what it is talking about.
Sometimes it does.
Sometimes it absolutely does not.
The problem is that both answers can sound equally confident.
We Are Laundering Expertise
Somewhere along the way, we started laundering opinions into expertise.
Someone with limited knowledge about a subject asks AI a question. The AI takes the assumptions contained in that question, combines them with an enormous amount of information and produces something that looks remarkably similar to professional analysis.
The idea went into the machine wearing sweatpants and came out wearing a suit.
That does not make the idea wrong. In fact, it may be excellent.
But the presentation can give the recommendation more authority than it deserves.
We at Divining Point have experienced this repeatedly over the past year.
Clients have sent us detailed marketing recommendations that were very clearly generated by ChatGPT. Sometimes they are presented as ideas for discussion. Other times they arrive closer to a list of demands.
We welcome the collaboration.
I would much rather work with a client who is actively thinking about their marketing than one who has completely checked out. Clients understand their businesses in ways we never will, and we understand marketing in ways they shouldn’t have to.
AI can make those conversations better.
It can also make them considerably worse.
The Recommendations Are Usually a Mixed Bag
This is what makes the problem so interesting.
The AI recommendations we receive are rarely complete garbage.
Some are things we are already doing.
Some are ideas we have already considered.
Some are ideas we advised the client to do years ago, only to have the idea rejected at the time.
Some are genuinely good suggestions worth exploring.
However, some would take the strategy totally off course.
The problem is, AI is right often enough that people have a difficult time recognizing when it is completely wrong.
AI Doesn’t Know What It Doesn’t Know About Your Business
This is especially important in fields where context matters.
Marketing is full of decisions that sound simple until you understand the business behind them.
Should you increase your Google Ads budget?
Maybe.
Should you publish more content?
Perhaps.
Should you target a broader audience?
It’s debatable.
Should you build more landing pages, change your messaging, target another market, lower your prices, increase your social media activity or completely restructure your website?
I’m not sure about that.
The answer depends on what you sell, who buys it, how they buy it, what you have already tried, what happened when you tried it, what your competitors are doing, how much capacity you have, what your margins look like, what your brand represents, what you are actually trying to accomplish, and on and on and on…
An AI model can only work with the context it has been given.
And people are notoriously bad at recognizing the context they have not provided.
This is an enormous blind spot that can turn perfectly reasonable advice into a remarkably terrible decision.
Sometimes AI Is Just Agreeing With the Question
There is another problem worth mentioning.
How you ask a question can influence the answer you receive.
Ask AI:
“Should my company create more landing pages?”
You might receive a thoughtful explanation of when additional landing pages make sense.
Now ask:
“My marketing agency has only created four landing pages this year. Would creating more landing pages improve our SEO and lead generation?”
Now you have introduced a premise.
Four does not sound like enough.
The AI may accept that framing and explain all the wonderful things additional landing pages could accomplish.
Twenty minutes later, someone forwards the response to their agency.
“We need more landing pages.”
But AI did not independently discover a weakness in the strategy. It responded to the way the problem was presented.
Every Professional Is Dealing with This Right Now
Doctors experienced the first version of this with WebMD. Now AI has put that phenomenon on steroids.
Lawyers will increasingly encounter clients who arrive with their own interpretation of the law.
Financial professionals will deal with people carrying AI-generated investment strategies.
Contractors will hear from homeowners who have already diagnosed the problem, selected the repair method and calculated what the job should cost.
Designers will be handed AI-generated concepts and asked to turn something that was never designed to actually work into a finished product.
And marketers will continue receiving 14-point strategies developed by a robot that has never spoken to a customer or attended a sales meeting.
We are entering a strange period where professionals are increasingly being asked to audit advice generated by AI for people who do not necessarily have the expertise to evaluate that advice themselves.
It is becoming a low-key digital terror campaign.
Information Is Not Experience
AI can give you information. But access to information is not the same thing as experience.
Experience is knowing which information matters in a particular situation.
Experience is knowing that something usually works, except when these three conditions exist.
It is knowing what happened the last five times someone tried it.
It is recognizing the difference between something that sounds strategically impressive and something customers will actually respond to.
It is understanding consequences that may not appear in the prompt.
Most importantly, experience includes failure.
Professionals become good at what they do partly because they have been wrong before. They have watched ideas fail. They have made bad assumptions. They have seen markets change. They have learned where the exceptions live.
AI can summarize those lessons, but it did not live through them.
Use AI to Ask Better Questions, Not Replace Expertise
None of this is an argument against using AI.
Quite the opposite.
AI may be one of the most powerful tools we have ever had for becoming efficient and better informed.
The distinction is how we use it.
There is a meaningful difference between asking:
“What should my marketing agency be doing?”
And:
“What questions should I ask my marketing agency about our current strategy?”
Ask AI what you might be overlooking. Ask it to argue against your assumptions. Ask it to explain an unfamiliar concept. Ask it what information would be necessary before making a recommendation. Ask it to give you arguments on both sides of a decision.
Then bring those questions into the room with the people who actually understand the context.
That is where AI becomes extraordinarily useful.