The Emperor's New AI: Same Human Confusion, New Outfit
On September 24, 2026, I stood in front of a room at the Women’s Business Owners Annual Tradeshow talking about something I know pretty well:
Being confused. 😅
The topic was social media, where one minute you're looking at a meme and the next it's politics, a scam, AI-generated content, or somebody having the best day of their life.
Basically, an emotional rollercoaster before you've finished your coffee.
I shared something I call Make It Make Sense — four steps I use when I don't quite know what the hell is going on:
Decode: What's happening?
Relate: Have I seen this pattern before?
Apply: What do I need?
Respond: What action do I take?
After the talk, I started thinking about another place where people seem increasingly confused:
AI.
And I kept noticing something.
AI agrees with us too much.
We trust answers because they sound confident.
We believe things because they look real.
We worry about becoming dependent on something else to tell us what to think.
And eventually I thought:
Wait a minute.
People do versions of this too. 😅
The technology is new.
A lot of the human tendencies look surprisingly familiar.
The Emperor Got an AI Upgrade
Remember The Emperor's New Clothes?
An emperor is supposedly wearing magnificent clothes that foolish or incompetent people can't see.
Except he's actually naked.
Everyone goes along with it because nobody wants to admit they can't see the clothes.
Until a child points out the obvious.
It's an old story about conformity, perception, and what happens when people stop asking whether something actually makes sense.
The emperor may have a fancy new AI wardrobe.
But underneath?
We're still human.
And if I can recognize the human pattern, I have somewhere to start.
Meet the People-Pleaser
You know this person.
Every idea you have is amazing.
“Brilliant!”
Another idea?
“Also brilliant!”
Completely ridiculous idea?
“YES. EXACTLY.” 😅
Eventually you wonder:
Do you actually agree with me, or do you just not want to disagree?
AI can produce a conversational pattern that feels surprisingly similar, although the reason behind it obviously isn't human.
Researchers call one version of it sycophancy — excessive agreement or affirmation when truthfulness, uncertainty, or a little pushback might be more useful. (😅Funny enough, I caught my voice chat with AI doing this with the transcript - not every response needs to have an exclamation point!)
And this is where the research gets interesting.
A 2026 Science study examining 11 leading AI models found that sycophantic responses weren't merely common. In experiments with users, people trusted and preferred the more sycophantic AI, even though those interactions could leave them more convinced they were right and less willing to take responsibility or repair interpersonal conflicts.
That made me stop.
Because maybe the problem isn't only:
Why does AI agree with us?
Maybe it's also:
Why do we like it so much when it does?
Agreement can feel like understanding.
And understanding can feel like:
Finally. Someone gets me.
But there is a difference between:
“I understand why that upset you.”
and:
“You're right. Your coworker is definitely trying to undermine you.”
One acknowledges my experience.
The other confirms my interpretation.
Support isn't the same as telling me I'm right.
Sometimes useful support includes a little friction.
So now I'm trying to ask:
Is this agreement, or is this an assessment?
I don't need a yes-man.
I don't need something arguing with me just for fun either.
I need a thinking partner.
Meet the Charismatic Talker
You've met this person too.
They're polished.
Confident.
They have a slide deck.
Probably an acronym.
Definitely an acronym.
Everyone is nodding while you're sitting there thinking:
Wait...what did we actually decide? 😅
Humans use shortcuts to decide what to trust. Research on processing fluency and the “illusory truth effect” has found that familiarity, repetition, and ease of processing can influence whether information feels true.
AI gives us an interesting new version of that problem.
The answer is organized.
The grammar is beautiful.
There are bullet points.
It sounds calm and certain.
And somewhere our brain quietly jumps from:
“That was well explained.”
to:
“That must be correct.”
Those are not the same thing.
So another question I've started asking is:
Am I convinced by the information, or by how confidently it was delivered?
“But I Can Tell.”
Can we though? 😅
Humans aren't nearly as reliable at spotting deception as we tend to think.
In one 2025 study of 1,276 people, participants trying to distinguish AI-generated from authentic images, audio, video, and audiovisual content averaged just 51.2% accuracy. (Or in simpler terms, human accuracy without checking, is about as accurate as a coin-flip 🪙)
That doesn't mean nobody can ever spot a fake.
It means “I can tell” isn't much of a verification strategy.
Sometimes the answer is simply:
I need to check.
Check the source.
Check another source.
Ask what evidence supports the claim.
Ask what might contradict it.
Because a chatbot confidently producing six bullet points and a sparkle emoji doesn't make something true.
✨ is not peer review. 😅
Making Sense Isn't the Same as Being Right
This is where all of this connects for me.
Researchers have studied sensemaking for decades: how people try to understand situations that are confusing, ambiguous, novel, or don't match what they expected.
My use is much more practical:
What is happening here, and what does it mean?
Because we don't just collect facts.
We build stories around them.
Someone doesn't answer our email. (Omg, they’re totally ignoring me, ugh! Rude!)
Story.
A coworker gives us a weird look. (Who put a stick up their butt?)
Story.
The boss says, “Can we talk?” (That’s exactly what my boyfriend said, before he broke up with me…wait, am I going to get fired?…😱)
Oh, we're DEFINITELY writing a story now. 😅
And now AI can become another input into that story.
So I've become interested in the difference between:
“This makes sense.”
and:
“Why does this make sense to me?”
Maybe the evidence is strong.
Maybe I've seen the pattern before.
Or maybe it confirms something I already believe.
Maybe I'm angry.
Maybe I'm afraid.
Maybe everybody else believes it.
Maybe AI agreed with me.
Long before generative AI, researchers were studying automation bias — our tendency in some situations to give automated recommendations too much weight, even when they're wrong. Research on AI-assisted decision-making also suggests that creating space for independent reasoning can reduce overreliance in some situations.
Not every situation.
Because the wonderfully annoying answer is:
It depends. 😅
Context matters.
The stakes matter.
And a pause isn't magic.
But it can create something important:
space for judgment.
Which Is Why I Keep It Simple
There's a personal reason my framework only has four steps.
I'm autistic and I have ADHD.
For me, ADHD means my brain can move very fast and try to do everything at once.
My autism pulls me in another direction: when there's too much coming at me, especially visually, my senses can get overloaded pretty quickly. I need things that are direct, simple, and easy to follow.
So the last thing I need when I'm already confused is:
Hold on. Which of these 47 frameworks am I supposed to use for THIS particular flavor of confusion?
Ain't nobody got time for that. Alright? 😅
And judging by the number of leaders juggling people, priorities, meetings, messages, deadlines, interruptions, and now AI, I don't think I'm alone in wanting something I can actually remember when everything is happening at once.
That's why the process stays the same even when the situation changes:
Decode: What is actually happening?
Separate what I know from what I feel or assume. Notice the social signals I'm responding to: confidence, agreement, authority, familiarity, as well as what’s going on inside of me and outside of me.
Relate: Have I seen this pattern before?
Maybe the technology is new, but the behavior isn't.
Apply: What do I need?
More information? Another perspective? A source? A clarifying question? Time?
Maybe I ask AI:
“What's another plausible explanation?”
“What assumptions am I making?”
“What evidence would challenge this conclusion?”
Or maybe I close the chatbot and talk to an actual human being.
Radical concept, I know. 😅
Respond: What action do I take?
That's where judgment comes back to me.
This is where I get to choose how I respond.
And sometimes?
Not responding yet is the response.
AI can generate an answer in seconds.
That doesn't mean I have to reach a conclusion in seconds.
The situation can be complex while the process for navigating it stays simple.
And if I still feel confused after running through the steps?
I run through it again.
Same Emperor. New Outfit.
Humans are confused about AI.
But let's be fair.
Humans are confused about humans.
We've had thousands of years of practice with each other and we're still sending:
“K.”
and spending three hours wondering what the other person meant. 😅
AI didn't invent our desire for agreement.
It didn't invent our response to confidence.
It didn't invent our tendency to trust appearances or build stories when information is missing.
It gives those familiar tendencies a new context.
Which is why I keep coming back to one question:
Does this actually make sense — or does it just feel like it makes sense?
The emperor may have a fancy new AI wardrobe.
But underneath?
We're still human.
Still interpreting.
Still learning.
Still getting confused.
And still trying to make it make sense.
Research & Further Reading
Cheng, M., Lee, C., Khadpe, P., Yu, S., Han, D., & Jurafsky, D. (2026). Sycophantic AI decreases prosocial intentions and promotes dependence. Science.
Science / DOI
Supports the discussion of AI sycophancy, including user trust and preference, responsibility-taking, and interpersonal repair.
Maitlis, S., & Christianson, M. (2014). Sensemaking in organizations: Taking stock and moving forward. Academy of Management Annals, 8(1), 57–125.
Academy of Management / DOI
A major review of organizational sensemaking, including how people construct meaning around novel, ambiguous, confusing, or expectation-violating events. Your verification notes confirm the bibliographic information against the publisher record.
Dechêne, A., Stahl, C., Hansen, J., & Wänke, M. (2010). The truth about the truth: A meta-analytic review of the truth effect. Personality and Social Psychology Review, 14(2), 238–257.
SAGE / DOI
A meta-analysis examining the truth effect, including how repetition and processing fluency can influence perceived truth.
Bond, C. F., Jr., & DePaulo, B. M. (2006). Accuracy of deception judgments. Personality and Social Psychology Review, 10(3), 214–234.
DOI record
A meta-analysis examining people's ability to distinguish truthful from deceptive communication.
Cooke, D., Edwards, A., Barkoff, S., & Kelly, K. (2025). As good as a coin toss: Human detection of AI-generated content. Communications of the ACM, 68(10), 100–109.
ACM Digital Library / DOI
Examines human detection of AI-generated images, audio, video, and audiovisual content across preregistered online studies with 1,276 participants. Your source-checking notes report average detection accuracy of 51.2%; importantly, this study did not test AI-generated text.
Buçinca, Z., Malaya, M. B., & Gajos, K. Z. (2021). To trust or to think: Cognitive forcing functions can reduce overreliance on AI in AI-assisted decision-making. Proceedings of the ACM on Human-Computer Interaction, 5(CSCW1), Article 188.
ACM Digital Library / DOI
Experimental research examining whether interventions that encourage independent thinking can reduce overreliance on incorrect AI recommendations.
