AI Doesn’t Read the Room
AI has been consuming everyone’s feed lately, and honestly, it should be. It has completely shifted the workforce, changed the way we communicate, accelerated content creation, and forced entire industries to rethink how work gets done.
When it comes to writing and communication work specifically, it’s impossibly to ignore the conversation happening right now around AI replacing roles tied to writing. And to be clear, I am not anti AI at all. I actually used AI often in my previous role for meeting notes, identifying process gaps, organizing revisions, version control support, and accelerating early drafts.
When used correctly, AI can absolutely be a useful tool.
But what AI still cannot replace is audience awareness.
That is something even the most experienced communicators spend years refining.
No matter how many systems AI is connected to or how much information it can process at once, it still does not truly understand the people receiving the communication. It does not understand emotional context, operational nuance, or the ripple effects communication can create once it reaches a real audience.
And I think that’s where a lot of this conversation gets oversimplified.
During a focus group at my previous company, we were testing ways AI could support revision tracking and version control. The issue was that the larger the scope became, the less reliable the results were. AI would miss revisions, interpret sections incorrectly, or create inconsistencies that required extensive human fact checking afterward. And I’m not talking about tiny errors, I’m talking major flaws that could create real operational risk if they were missed.
The most effective solution I found was to narrow the scope intentionally. Instead of asking AI to compare entire documents at once, I started breaking revisions down section by section so the output was smaller, easier to validate, and less risky to review. Internally, it eventually became jokingly referred to as “The Mariah Method”, which I still laugh about. At one point, even AI started referring to it as “The Mariah Method” in the meeting notes.
But honestly, that experience reinforced something important for me:
AI works best when experienced people know how to guide it.
Not when it replaces them entirely.
There are also certain areas of communication work where I personally would never rely heavily on AI, especially escalation call scripting and emotionally sensitive customer communications. AI tends to interpret communication literally, but experienced communicators understand that people rarely receive communication literally. They receive it emotionally.
The same sentence can calm someone down, frustrate them further, or escalate an already difficult interaction depending on tone, wording, timing, and audience awareness. That level of judgment comes from years of operational exposure, collaboration, listening, revising, testing messaging, and understanding how real people react under stress.
Good communication is not just accurate. It’s aware.
And I think that’s the part people underestimate most about experienced communication work. The real value is never just producing words quickly. It’s understanding what happens after those words reach real people.