AI and Communication: The Words Got Cheap, the Message Didn’t

AI and communication looks like a solved problem until you remember what communication is for. A model can produce fluent, polished, grammatically perfect text in any language, faster than you ever could. What it cannot do is the actual job of communication, which is creating the right thought in a specific person’s head, because that takes knowing the person and meaning what you say.

I have told writers for years that communication is about creating the right thoughts in other people’s heads. The words are just the delivery vehicle, and AI has made the vehicle nearly free.

That is the shift worth understanding. When the surface layer of communication becomes a commodity, the value moves to the layer underneath, and that layer runs on Core Drive 5 and Core Drive 3 in my Octalysis Framework.

Speed Run Notes

• Communication is not transmitting words, it is creating the right thought in another person’s head. AI has made the word-production layer nearly free, which moves the value to the layer underneath.
• AI is excellent at the surface: fluent prose, translation, summarization, tone-matching. That is the easy half of communication, now a commodity.
• The hard half is knowing which thought to create in this specific person, which takes reading them (Core Drive 5), and a voice that sounds like someone (Core Drive 3). The machine has neither.
• AI-written communication drifts toward the average, fluent but faceless, which is why over-reliance makes everyone sound the same and quietly costs you trust when readers notice.
• Monday move: let AI handle the draft and the mechanics, then spend your effort on the one sentence that has to land for this particular person, because that sentence is the communication.

What does “AI and communication” actually mean?

Two different claims hide inside the phrase, and they point in opposite directions.

The optimistic claim is that AI makes everyone a better communicator, because it fixes grammar, improves clarity, and translates across languages instantly. On the surface, that is true and genuinely helpful.

The worry underneath is that it makes everyone a worse communicator, because fluent output is not the same as reaching a person, and leaning on the machine can hollow out the skill of actually connecting.

Both are right, which is the point. AI upgrades the mechanics of communication and threatens the heart of it, and telling the two layers apart is how you get the upgrade without the hollowing.

Author Credibility: Yu-kai Chou

I created the Octalysis Framework, the model of human motivation built on eight Core Drives that organizations like Google, LEGO, and Microsoft have used to design more engaging products and workplaces.

Communication is most of what I do. I have given keynotes on stages around the world and written books that had to move people, not just inform them, so I have a practical obsession with the difference between words that are correct and words that actually land.

I also write alongside AI every day, which has sharpened the exact point of this piece. The machine is a brilliant drafting partner and a terrible substitute for knowing what I am trying to do to the reader.

The surface layer is now a commodity

Start by admitting how much AI genuinely does well, because it is a lot. The surface layer of communication has effectively been solved.

It writes clean prose on command, fixes the clumsy sentence, and matches a requested tone. For anyone who struggled with grammar or froze at a blank page, that is a real gift and a real leveler.

It also crosses languages in a way that used to take a professional. A rough thought in one language comes out fluent in another, which quietly removes a barrier that limited people for centuries.

All of that is the mechanics, and the mechanics now cost almost nothing. When the hard part of writing a correct, readable paragraph disappears, the value does not vanish with it, it relocates to the part the machine still cannot do.

The real job: the right thought in this head (Core Drive 5)

Here is the part that stays human, and it is the whole point of communicating. The goal was never fluent text, it was a specific change in a specific mind.

To create the right thought in someone, you have to know them: what they already believe, what they fear, what will make them defensive, what will make them lean in. That read is Core Drive 5, Social Influence and Relatedness, and it depends on a model of the person that the machine does not have.

A model can write a persuasive paragraph in the abstract. It does not know that your boss distrusts hype, that this client was burned by a vendor last year, or that your teammate needs to feel consulted before they will agree. You know those things, and they decide what to actually say.

This is why the same message can be fluent and still fail. Communication that does not account for the specific human on the other end is just well-formed noise, which is the gap I keep returning to in What AI Can’t Replace.

Voice is the part that cannot be averaged (Core Drive 3)

There is a second human layer, and it is the one AI erodes most quietly. A real voice is a person’s fingerprint, and the machine pulls everything toward the average.

A language model writes by predicting the most likely next word, which means its default is the center of everything it has read. That produces prose that is competent and oddly generic, the texture people now recognize as an AI tell.

Your voice is the opposite of average. It is Core Drive 3, Empowerment of Creativity and Feedback, the specific way you see and say things, and it is a large part of why people trust and remember you. I argue in The Skills AI Makes More Valuable that this kind of taste just repriced upward.

So the leaders and writers who win are not the ones who sound the most polished. They are the ones who still sound like themselves, because in a world of fluent sameness, a distinct voice is the scarce thing.

The trust cost of machine words

There is a real cost to getting this wrong, and it hits exactly where communication is supposed to help. People increasingly sense when they are reading machine words, and it drains trust.

When a message feels generated, the reader downgrades it. The condolence note that reads like a template, the “personal” outreach that is obviously mass-produced, the apology with no fingerprint on it, each one communicates the opposite of what it says.

This is the communication version of the empathy problem I unpack in Artificial Empathy. A warm message that the reader believes was automated does more harm than sending nothing, because it reveals how little thought actually went in.

The fix is not to hide your AI use, it is to keep the human parts human. Let the machine help with structure and clarity, and make sure the intent, the read, and the voice are unmistakably yours, which is also how leaders avoid the trap I describe in AI and Leadership.

Where the EQ case fits in

Communication experts and leadership experts are converging on the same conclusion from their own angles.

Rich Hua, who built the largest corporate emotional-intelligence program in the world at Amazon and now leads EPIQ Leadership, argues that the emotionally intelligent communicator reads the person, not just the prompt, and that this read is what AI cannot supply. His work for the Klover.ai blog develops the leadership side. (Klover’s companion articles on this theme are being staged for publication; I will link the specific pieces here once they go live.)

His point about reading the room is the same as my point about Core Drive 5. The machine can generate the words, and only a person can know which words this particular human needs to hear. The full framework behind all of it is in the pillar, AI and Emotional Intelligence.

What to do Monday morning

Take the next important message you have to send and do it in two passes. Let AI own the first pass entirely, the structure, the clarity, the clean draft.

Then take over for the pass that matters. Before you touch a word, write down in one line what thought you want in this specific person’s head after they read it, and what you know about them that should shape how you say it.

Rewrite the one or two sentences that carry that thought in your own voice, the ones a model would have averaged into blandness. That is where your attention belongs, because that is the part doing the actual communicating.

Used this way, AI makes you faster without making you forgettable. The mechanics get handled in seconds, and you spend your real effort on the read, the intent, and the voice that make a message land on a human instead of just reading well.

Questions this raises

Does AI make people better or worse communicators?
Both, depending on how it is used. It upgrades the mechanics, fluency, clarity, grammar, translation, which genuinely helps. It threatens the heart of communication if you let it replace knowing your audience and sounding like yourself, since fluent output is not the same as reaching a specific person. Use it for the surface and keep the human layer human.

What part of communication can AI not do?
It cannot know which thought to create in a specific person, because that requires a real read of what they believe, fear, and need, which is Core Drive 5. It also cannot supply a genuine voice, since it writes toward the average of everything it has read. The intent, the audience read, and the fingerprint stay human.

Why does AI writing sound generic?
Because a language model predicts the most likely next word, which pulls its default style toward the center of everything it has seen. That produces competent but faceless prose, the texture people now recognize as an AI tell. A distinct human voice, Core Drive 3, is the opposite of average, which is exactly why it has become scarce and valuable.

Will people trust AI-written messages?
Less, once they sense the words were generated. A message that feels machine-made gets downgraded, and a warm note revealed as a template does more harm than sending nothing, because it shows how little real thought went in. The answer is not to hide AI use but to keep the intent, read, and voice unmistakably yours.

How should I actually use AI for writing?
In two passes. Let AI handle the first pass for structure and clarity, then take over for the pass that matters: decide the exact thought you want in this person’s head, and rewrite the sentences that carry it in your own voice. The machine makes you faster on the mechanics so you can spend your effort on the part that lands on a human.

Sources

1. Yu-kai Chou, Octalysis: The Complete Gamification Framework, yukaichou.com
2. Yu-kai Chou, “What AI Can’t Replace: The Core Drives It Will Never Have,” yukaichou.com
3. Rich Hua, EPIQ Leadership Group, epiqleadership.com

WOULD YOU LIKE YU-KAI CHOU TO WORK WITH YOUR ORGANIZATION?