Thinking Techniques to Avoid Being Deceived: Why People Who Speak Fluently—in Meetings, Lectures, and ChatGPT—Are the Most Dangerous

"This person is so smart!"
"This teacher's lecture is so easy to understand!"

The moment you think that, in reality, your brain might be getting deceived.

Humans have a habit of "trusting things that are fluent."
That is the fluency heuristic.

This time, a few technical terms (with explanations) will appear, but I'm sure you'll be convinced once you finish reading, so I hope you'll do your best to keep up.

Actually, just humor me and try reading it! 😇

"Fluent = Correct" might be an illusion

When I was studying for the SME Consultant exam, I was once overwhelmed by the way a certain instructor spoke. I was stunned by their fast, smooth delivery and the way they used technical terms with such rhythm, thinking, "This person is amazing..."

However, when I later studied that field deeply on my own, I realized that they had been confidently stating things that were clearly factually incorrect or simply their own subjective opinions.

At that moment, I finally understood.
"They were just good at speaking."

I had entrusted my judgment not to whether the content was correct, but to how "convincing" the delivery sounded. This is the true nature of the fluency heuristic.

What is the fluency heuristic?

In the world of psychology, it is defined as follows.

People tend to judge information that is processed smoothly as "correct." Easy-to-read materials, clear language, and a smooth tone. Such "fluency" allows even the validity of the content, which should be checked, to slip through.

Meetings, presentations, lectures, and recently, ChatGPT. All of these are full of room for this bias to creep in. Don't you think that's very scary?

That person who speaks fluently isn't necessarily prepared

I have seen several instructors who talk about being able to give a lecture with zero preparation as being "professional." Certainly, they may have a certain level of knowledge. However, that does not mean it is being conveyed correctly to the students.

Rather, behind the culture that appreciates such "improvisation" and "no-plan ability" is an unconscious assumption that "if it's fluent, it must be correct.".

On the other hand, we instructors are inevitably required to have "improvisational skills." This is because, in many cases, a Q&A session is scheduled at the end of a seminar. The instructor needs to grasp the intent of the participant's question, construct an answer on the spot that is consistent with the seminar content, and convey it clearly. The time given for that thinking is only a few seconds. Therefore, when you work as an instructor, you acquire "improvisational skills." This

"improvisational skill" can be a weapon, for better or worse.

The "plausibility" lurking in ChatGPT as well

I use ChatGPT every day for work myself. It is convenient for creating drafts of content and brainstorming ideas. Creating reports and meeting minutes has also become much easier.

However, I sometimes feel a sense of discomfort, like, "Isn't this a bit too well-summarized?"

For example, when I ask for an overview of a subsidy program, it answers in text that sounds correct. However,
• information for the latest fiscal year is missing,
• it is ambiguous who the support is for, or
• the essential perspective of practical application in the field is lacking.

Even if it looks perfectly fine on the surface, it is actually full of holes. This, too, is a fluency bias.

Of course, it can be manipulated in any way depending on the prompt (the text used to give instructions), but the person asking the questions also needs a certain level of literacy.

Without symbol grounding, words are just decoration.

So, why are we deceived by this "meaningless fluency"?

The concept behind this is symbol grounding Have you heard of it?

What's that? I've never heard of it!
Don't think it sounds too difficult...

This is where it gets interesting! 😄

Symbol grounding is the idea that words (symbols) only acquire true meaning when they are connected to "physical sensations" or "experiences".

In Mutsumi Imai's book, "Cognitive Psychology for Correctly Facing Life's Big Problems," it is stated as follows:

For example, when you hear the word "melon," the sweet scent that comes to mind, the rough surface, the flavor spreading in your mouth—that is "grounding." When a baby learns the word "doggy," they only truly understand "is this a dog?" after actually touching one. When you hear "snow" and imagine the biting cold, or recall the sensation of it melting away on your hand, that also comes from real-life experience.

On the other hand, the text generated by AI lacks this "grounding". It is simply arranging words plausibly based on past data, and it does not feel the meaning.

Beware of politicians who dodge the point

I like watching political content, so I often watch regular press conferences of the Cabinet and party leaders, as well as debate programs before elections.
(My favorite channel is "ReHacQ." I've liked Producer Takahashi since his TV Tokyo days.)

However, in debates between parties, many politicians dodge the point without answering the question and speak in a way that "sounds plausible." They use makeshift phrasing not to convince the listener, but to silence them. It's a common occurrence.

As a listener, I want to retort, "That's not an answer! You'd get a zero on the management consultant exam!"

They, too, are dangerous beings who "do not have grounded symbols" but "are fluent".

Three perspectives to avoid being deceived

So, how should we confront this combo of fluency bias and lack of symbol grounding?

I will introduce three methods that I practice.

1. Have questions that connect to your experience
"Does this connect to my own experience?"
"What would happen if I used this myself?"
By passing information through these questions, you can prevent the illusion of understanding.

In my lectures, I explain things by using familiar examples and by conveying that even abstract phrasing encountered when solving problems should be grasped as concretely as possible. This is precisely because being conscious of "symbol grounding" increases the correct answer rate.

2. Rephrase in your own words
Whether it is text obtained from ChatGPT or content heard in a lecture, you cannot master it unless you can convert it into your own words. If the words are not "originated by you," they are still someone else's knowledge.

3. Try explaining it to a third party
Fluent words make you feel like you have "gotten the message across." However, when you try talking to a third party, you often feel that "I can't explain this well." That is where a "lack of grounding" in understandingappears.

What I wanted to address in this article were the keywords "fluency bias" and "symbol grounding." By explaining these, my goal was to ground them within myself.

Let's reclaim the "meaning" of communication in the AI era!

As AI becomes commonplace, presentations prioritize appearance, and language becomes increasingly refined. The same goes for writing. There are quite a few articles on this Note platform that are clearly written by generative AI (I don't deny this, as it is not a bad thing).

Mutsumi Imai warns us of this:

This phrase really resonated with me.

In our pursuit of "efficiency," we must not forget the human activity of "grounding."

Truly valuable communication is not about fluency.
It is about whether words are connected to experience.

Things you can talk about and convey through your own experience. That is where human "strength" and "the power to connect" reside. That is why I will continue to write articles in my own words.

And literacy as a reader, this is also important.

It was a great book that also taught me "What is cognitive psychology?" 👍

If you found this interesting, I look forward to your "likes" and "shares"!

Click here for more on this Note
↓