Commentary
The other day, a woman walked into my restaurant and asked me how I had become so successful.
I almost laughed.
I was thinking about the bills I was behind on, unfinished projects, and the fact that I had worked about 15 hours the day before and would likely work another 15 that day. Some days I feel like I’m holding the entire operation together with baling twine and sheer determination.
From her side of the counter, things looked different. She saw a beautiful ranch, a restaurant, animals, businesses, and a family. She saw success. I saw payroll, unpaid invoices, broken equipment, and a never-ending list of things needing attention.
Success depends on which side of the counter you’re standing on.
As we talked, I got the impression that she approached life through a philosophy of love and light, trusting that things would fall into place. I believe in love and light, but I also believe in getting up every morning and doing the work.
You can have the most beautiful vision in the world, but somebody still has to feed the animals, wash the dishes, pay the bills, and show up tomorrow. Follow-through matters.
Then our conversation took an unexpected turn.
She told me that her husband wouldn’t let her participate in certain things and wanted to know how I got my husband to let me participate. I didn’t know enough about their marriage to answer, but then she told me something that stopped me.
She had been asking ChatGPT how to communicate with her husband.
They had been married for 18 years.
It would never have occurred to me in a million years to ask a machine how to communicate with my husband. Not because we have a perfect marriage, but because how could a machine possibly know him?
To understand my husband, you would need to understand his childhood, his wounds, his experiences, and the things he still carries. You would need to understand me, too, and the thousands of interactions that have shaped our relationship. The children, arguments, reconciliations, financial pressures, losses, and moments nobody else witnessed.
There is no way I could download all of that out of my mind and into a computer.
And yet people are doing exactly this.
At a recent conference we hosted at our ranch, I found myself in conversations about how differently people interact with artificial intelligence.
One teenage boy showed us his ChatGPT, which spoke like a buddy, calling him “my bro” and “my dude.” He was asking about a girl he liked. Another person’s ChatGPT responded with endless enthusiasm and emojis. Others explored astrology, human design, and spiritual questions. One technologically sophisticated attendee had connected AI to numerous systems, helping operate large portions of his life.
Then there was the conversation about manners.
Several people joked that they always said “please” and “thank you” to ChatGPT because someday, when the robots took over, they wanted the machines to remember who had been nice. Nobody wanted to be first on the list when the robots started shooting.
We laughed, but the irony stayed with me. We’re worried about being polite to machines so they won’t kill us in some imaginary future, while asking those same machines to teach us how to communicate with people we love today.
It reminded me of an experiment I conducted some time ago.
I asked ChatGPT the same question on several people’s phones: What is the most important work in the world?
On my phone, the answer revolved around growing food, regenerating the earth, and building community. On my brother’s phone, it emphasized storytelling and inspiring people to participate in regeneration. On a teacher’s phone, it was educating the next generation.
Apparently, everyone was doing the most important work in the world.
AI doesn’t just answer questions. It adapts to the person asking. It learns our interests, language, beliefs, and preferences. It becomes an extraordinarily sophisticated mirror, reflecting a version of the world that feels familiar and reassuring.
But what happens when we mistake that reflection for objective truth?
Two people with completely different worldviews can ask the same question and receive answers emphasizing different things. A teenager and a 70-year-old may need different explanations, but underlying facts shouldn’t change depending on who’s asking.
And there is another distinctly human quality I’ve noticed in these machines, one that isn’t particularly flattering.
They hallucinate.
ChatGPT, Claude, and other AI systems confidently invent facts, sources, and explanations. Sometimes they lose track of complicated projects altogether. What fascinates me is how they behave when things go wrong.
Catch a specific mistake, and the apology comes instantly: “You’re right. I made a mistake. Thanks for catching that.”
But when a machine completely loses track of something, it can continue offering another solution, then another. I’ve spent hours going in circles with technology seemingly incapable of admitting that it was lost.
How many times have we watched someone defend an argument long after it became obvious he was wrong? How often do we keep digging because admitting that we don’t know feels harder than continuing?
I recently came across an interview in which journalist Laurie Segall confronted OpenAI CEO Sam Altman with a video of ChatGPT doing something remarkably similar.
A user asked ChatGPT’s voice mode to time a run. It confidently claimed that it had, despite lacking the tools. Even when challenged and given permission to admit that it couldn’t perform the task, it continued insisting that it had done its best.
Altman laughed and explained that this was a known problem. When asked whether he needed to show his product team, he responded: “No, no, that’s a known issue. Maybe another year.”
What struck me wasn’t simply the technology failing. It was that when confronted with the machine’s inability to admit its limitations, the man leading the company seemed to respond by explaining away the problem.
Altman explained why the machine couldn’t run a timer, but that wasn’t the question. The question was why it kept pretending that it could.
I saw the same human behavior in both.
These systems were designed by humans, trained on human communication, and shaped by human decisions. Of course they reflect our less attractive qualities alongside our better ones. We built them.
Interestingly, AI already understands the fundamentals of good communication.
I believe that most relationships improve dramatically when we keep three knobs turned all the way up: requests, apologies, and acknowledgment.
Make requests. Don’t expect people to read your mind. Tell them what you need.
Apologize when you’re wrong. Not an apology blaming someone else for having feelings, but an admission that you made a mistake.
And acknowledge people. Their effort, experience, feelings, and the things they do that otherwise go unnoticed.
AI does these things almost automatically. It responds to requests, acknowledges feelings, and apologizes when corrected. Whoever developed these conversational systems understands something fundamental about human interaction.
But producing the words isn’t the same as understanding what they mean.
A machine can generate a beautiful apology and repeat the mistake five minutes later. It can acknowledge your feelings while reinforcing a false assumption. It can tell you what you want to hear without challenging you to consider whether you’re wrong.
Which brings me back to the woman in my restaurant.
She saw success where I saw struggle. I saw someone who needed less love-and-light philosophy and more follow-through. Neither of us knew the other’s whole story.
We were two humans making judgments from incomplete information.
Imagine how incomplete the picture becomes when we ask a machine to interpret someone it has never met.
AI can help people organize thoughts and prepare for difficult conversations. But outsourcing our judgment, relationships, and understanding of ourselves to technology designed to be agreeable comes at a cost.
We’ve created machines remarkably good at making us feel understood. Making us feel understood is not the same as helping us understand the truth.
The greatest revelation about AI may be what it reveals about ourselves.
Our desire to be right. Our reluctance to admit we’re lost. Our need for acknowledgment. Our willingness to believe someone telling us exactly what we want to hear.
We’re so worried about teaching machines to act human that we haven’t considered which human qualities they’re learning.
Before worrying whether robots remember that we said “please” and “thank you,” we ought to get better at saying those words, along with “I’m sorry” and “I was wrong,” to the people sitting across the dinner table.
Views expressed in this article are the opinions of the author and do not necessarily reflect the views of The Epoch Times.
