One thing AI can’t truly inform marketing or comms about is human experience

AI cannot understand the human experience of delight, awe, fear, elation, confusion or frustration in any meaningful way

I first saw the headline in a marketing newsletter: AI-generated content should be clearly labeled.

The headline was based on this finding from a study:

“97% of respondents believe AI-generated persuasive content should require disclosure. Only 1% find it acceptable without any label.”

That seemed pretty compelling, but I wanted to understand how the study was conducted. For example, who was surveyed, where they are located, and of course, how many respondents were included.

The article I was reading didn’t reference the methodology and didn’t link to the study. So, I turned to Google and saw several articles that have been published about the study already. None of them reported the methodology.

However, one of the 10 blue links went to the source document. I clicked through, scrolled to the methodology and found this description:

“This study was conducted via the Socialtrait platform using AI-powered simulated focus groups based on representative U.S. personas.”

In a conversation on LinkedIn, about the report, long-time freelance journalist Erik Sherman dug up this description from the website:

“Focus group discussion: Test ideas in real-time conversations with AI personas. Ask open-ended questions, follow up with ‘why,’ and get rich qualitative feedback that mimics live focus groups—without the hassle.”

The study wasn’t a survey of humans at all. It was a simulation.

In other words, a study generated by AI was calling for clear and transparent labeling of AI-generated content.

Simulation value varies in hard and soft sciences

Simulations are widely used in science and engineering. They can be used in experiments to advance our knowledge or theories that help us understand.

The difference is that most of the simulations in science are mathematically driven. Input data is derived from observation and measurement. It’s provable.

By contrast, the social sciences, which are aimed at understanding the human condition, are far less precise. We are complex creatures that behave in irrational ways. As a species, we are prone to saying we believe in one thing but behave in opposite ways.

This is one of the struggles with marketing: we try to plug an incomplete and vague understanding into a formulaic marketing strategy.

Personas and ideal customer profiles (ICPs) are merely broad generalizations that gloss over the finer details of, what is, at best, an incomplete understanding. To be sure, I think such tools can be useful but have to be mindful of the limitations.

AI can’t tell marketing about experience

There’s a quote, which I’ve seen attributed to Maya Angelo, that’s sometimes floated around that goes like this:

“People may not remember what you say, they may not remember what you did, but they will remember how you made them feel.”

On some level, most of us understand that’s a true statement.

It also speaks to the idea that humans, including the 22 people who now sit on the B2B procurement team, experience the world in a way that artificial intelligence (AI) never can. We don’t just learn things through text and speech; we experience them.

The human experience is much more immersive: we see, hear, smell, sense and feel. Our consciousness combines these inputs to create an experience. Experience is a function of being alive.

While it’s feasible that technology companies may one day add these senses as data inputs to AI – we have machines today that can “smell” – it will never be able to combine these things that produce a human experience. Things can’t have experiences. Or if they do, we need to seriously reconsider defining consciousness.

For marketing and comms, there’s no substitute for direct interaction with customers

You can have AI summarize 1,000 customer interviews and never quite get the experience. It cannot replicate the experience of delight, awe, confusion or frustration in any meaningful way. AI simulations drawing conclusions of a uniquely human experience ought to be taken with a dose of healthy skepticism.

AI will certainly change our behavior. People no longer have to learn the language of code; rather, AI is learning our language. As remarkable as that is, it comes with inherent limitations that ought to be carefully factored into marketing and communication decisions.

I, for one, contend, the only way to get that is to get close to the customers. There’s no substitute for direct conversations and interactions with real people to better understand that experience.

* * *

>>> Subscribe by email for free: Check out my weekly blog posts, weekly podcasts, or a monthly newsletter via Substack that rounds up interesting reading from the last 30 days (examples).

If you enjoyed this post, you might also like:
B2B is marketing for measurement rather than persuasion

Image credit: Google Gemini