Things I Wish Someone Had Told Me About AI Before My First Communications Job
By Marya Kalra
When I started my first communications role, I thought knowing how to use AI made me prepared. I could prompt it, edit it, and get a decent first draft in under two minutes. I figured that was the skill. Turns out, it was only half of it.
The other half is knowing when not to. And that’s the part nobody teaches you.
For my capstone thesis at Humber, I spent months researching exactly this: how employees and organizations actually experience AI in internal communications. I surveyed working professionals and sat down with senior communicators from banks, PR agencies, retail companies, and consultancies. What I heard changed how I think about AI at work. Not whether to use it — but where it genuinely helps, and where it quietly makes things worse without you realizing.
Here’s what I wish I’d known going in.
Feeling comfortable with AI is not the same as knowing when it’s right.
This was the biggest thing that shifted for me. There’s a real difference between being good at using a tool and having the judgment to know when it belongs in the room.
I see this in myself and I see it in a lot of people early in their careers. We learn the tools fast. We get efficient. And then we start applying them everywhere, because why wouldn’t we? But some messages aren’t about efficiency. Some messages are about a person on the other end who’s worried, or confused, or going through something hard — and what they need isn’t a well-structured paragraph. They need to feel like a human being actually thought about them.
That’s where AI, on its own, consistently falls short.
Know which moments are off-limits. Seriously, know them.
Every senior communicator I spoke with for my research, without me asking them to, said some version of the same thing: there are communication moments where AI just doesn’t belong. Crisis situations. HR conversations. Layoffs, restructuring, performance messaging. Leadership communications that carry real emotional weight.
One person described those moments as ones that “call for authenticity and sincerity that AI cannot capture.” And they were right. These aren’t moments where being fast or polished is the point. They’re moments where the person reading the message needs to feel like someone actually showed up for them. AI doesn’t do that. It produces text. There’s a difference.
Early in your career, it’s worth building a clear mental list of the communication types where you keep AI out entirely. Not because you can’t use it — but because the stakes of getting the human element wrong are too high.
Use it to start, not to finish.
The best way I’ve heard AI described by people who actually use it well is as a thought partner. Something to bounce off, get unstuck with, get a rough shape from — and then you take over completely.
The mistake I made early on was treating a decent AI draft as a near-final product. The edits I made were surface-level. And the result was copy that was technically fine but felt a little… hollow. A bit generic. Like it could have been for any company, any audience, any moment. Because it was.
Your job as a communicator isn’t just to produce words. It’s to make people feel something — informed, included, reassured, energized, whatever the moment calls for. AI can give you a structure. It cannot give you that, it simply does not have the emotional depth for it, only you can. Which is why one of my research findings was that employees do not trust AI, they trust the human supervising AI.
People can tell. More than you think.
Something that surprised me in my research was how often employees said they could already sense when a message had been AI-generated — and how that affected the way they read it. Not always consciously. Just a feeling that something was a little off. A bit flat. A little too smooth.
Your audience has good instincts. And in internal communications especially, where trust is everything, a message that feels manufactured can quietly chip away at exactly the thing you’re trying to build. It doesn’t take a big scandal. It’s just a slow erosion of “does leadership actually care, or are we just getting content.”
The fix is simple but it requires actual effort: read your draft out loud before you send it. Does it sound like a person? Does it sound like your organization? If the answer is no, keep editing until it does.
The best communicators aren’t the ones who use AI the most. They’re the ones who know when not to.
This is the thing I keep coming back to. We’re entering the field at a moment where everyone is figuring out the rules in real time. There are no perfect playbooks. Most organizations are genuinely still working this out.
That’s an opportunity. The communicators who are going to be worth something in this environment aren’t the ones who can automate the fastest. They’re the ones who can look at a situation, think about the person on the other end, and make a real judgment call about what that moment needs.
Your judgment and your empathy is the thing AI can’t replicate. Protect it. Develop it. And use it — especially when it would be easier not to.
Marya Kalra is a recent graduate of the Bachelor of Public Relations (Honours) program at Humber Polytechnic and currently works as a Communications Coordinator at Martin Brower of Canada.