4 leadership capabilities for communicators in the AI age.
I get asked a lot what capabilities communicators and marketing leaders actually need to lead in an AI world. I’ve put a lot of thought into this matter and tested ideas and models with academic and professional audiences.
The model has four elements:
Data.
AI literacy.
Business processes.
Change management.
Often, organizations start with the second one. They start with “Subscribe to some AI tools. Then teach people to use those AI tools.” I think that’s backwards.
I think every AI story starts with data, even if you are using it for a creative purpose like writing or image/video composition. When I say data, I don’t mean that you must become a data scientist, or even that you should use quantitative data. Rather, I am saying that you should organize your sources, label them and group them in a folder or database because both you and the AI you are using need to know what’s real. What data exists, who owns it, what quality it’s in, what’s sensitive, what you’re allowed to use. And most importantly, what decisions are data supposed to inform? If you don’t have that, AI won’t create insight; it will just deliver fluent, confident-sounding slop.
Second is AI literacy. Notice I’m not just saying “prompt engineering.” Being a prompting expert is a skill that every professional will need to develop, but first they need to develop judgment. What’s this AI tool good for? What’s risky? Where does it need human review? Where does policy draw a line? You’re building the mental model of your workflows, so that you can rebuild them around AI. It’s the opposite of a bag of tricks, really. This kind of thinking is strategic and systematic - a review of your processes to see what parts the particular AI tool can transform.
Third is business processes. This is where most AI efforts fall apart. AI makes you and your team more effective by changing how work moves. You should be able to answer questions like these: “Okay, where does AI touch this workflow? Why did you choose this AI tool (AI literacy)? Who reviews? What’s the escalation path? What’s the definition of done?” And where does the source of truth live (data)? If you can’t answer those questions, AI just turns into random drafts in random places, which wastes time. In my observation, a lack of understanding and thorough documentation of business processes is one of the major reasons that AI pilots in communications fail.
Fourth is change management. To be honest, in my experience, most AI pilots fail for human reasons. People don’t trust the technology. They hesitate to use it because they don’t know what’s allowed. Or they think it’s a surveillance tool. So your job is to make the guardrails explicit by: (i) training people in plain language; (ii) running small pilots. (iii) making regular review and reporting of AI use a habit. Also, creating a community of practice where people are invited to be open and vulnerable: “If you’re unsure, ask.”
Quick Case Study
Say a hospital foundation wants AI to speed up the drafting of donor thank-you letters.
Let’s consider data first: What’s permissioned? What’s sensitive? What should never go into a tool?
Next, AI literacy. Pick approved uses, like drafting from public campaign facts, not storing donor data, not making claims without sources.
Business Processes: map the steps. Intake. Draft. Human review. Approval. Send. Log. Make review non-optional.
Change management: frame it as standardizing quality and reducing busywork, not as surveillance. Train on good and bad examples. Measure time to first draft and revision count, not volume.
There you have it. A quick example of how to implement the model.