The real AI strategy is giving people room to try

I have absolutely used AI to get unstuck.

I have ADHD, and my executive function sometimes works like a ‘92 Mazda 626 (my actual first car) trying to start in the dead of winter in Saskatchewan–a bunch of false starts with nothing to show for it. So when an email slides into my inbox and requires a lot of detailed thought, I often use AI to help organize my ideas, summarize my long-winded, too-much-context thoughts, and put together a first draft that is the metaphorical boost my executive function needs to thrive. 

So this is not an anti-AI note. (That, frankly, would be strange for Growclass to write, given that we teach AI strategy and spend a lot of time helping folks figure out how to use these tools…😅)

But if you read enough online, scroll enough LinkedIn posts, open enough newsletters, review enough partner emails, or consume enough content that was very clearly written by someone asking AI to “make this sound more human,” you start to notice something.

You can spot the AI voice. (Or, maybe more like, the lack of a point of view and personality)

There is no real texture. No weirdly specific detail that makes you think, “Ah, a person wrote this.” No sense that someone sat with the idea long enough to figure out what they actually believe. It is content at the speed of convenience, and sometimes convenience is great. But sometimes, convenience is actually just more beige internet slop.

And for anyone trying to build trust with an actual audience of actual humans, that matters.

Writing is one of the ways people decide whether you understand them. It is how they decide whether they trust you. It is how they decide whether your organization knows the inside of their problem or whether you asked a tool to “make this sound warm and professional” and then called it a day.

And let me tell you the hard truth: people are smart, and they can tell.

This is why we wrote the op-ed

Recently, Growclass and Women in Communications and Technology wrote an op-ed for The Hill Times about Canada’s AI For All (cute name! 😜) strategy. We cannot republish the full piece just yet, but you can read it here!!

The argument of the piece is pretty simple: Canada’s AI strategy gets a lot right, but strategy is not the same as execution.

Canada helped build the global AI ecosystem. We invested early in research, developed world-class institutions, and contributed talent that shaped the technology’s evolution. That is all worth being proud of.

But leadership in innovation is not the same as leadership in adoption.

That distinction matters because AI adoption is not just a technical issue. It is a trust issue, a training issue, a workplace culture issue, and a participation issue. It is about whether people know how to use these tools in their actual work. It is about whether they trust the systems enough to try. It is about whether employers provide meaningful training, instead of assuming people will figure it out on their own between meetings. And it is about who gets invited to build, lead, question, and shape what comes next.

This is where we need to be careful about how we talk about the “confidence gap.”

When women are less likely to use AI tools at work, the easy explanation is often that women are less confident, less interested, or more hesitant. But that framing misses the bigger picture. Confidence does not exist in a vacuum. It is shaped by access, culture, bias, and risk.

If someone has fewer opportunities to learn, fewer people informally showing them how tools are being used, fewer leaders encouraging experimentation, and a higher chance of being judged harshly for making a mistake, lower adoption should not surprise us. That is not a personal failing. That is a system working exactly as designed.

This is especially important for women and underrepresented people, who are too often included in AI conversations only through the lens of risk. Harm, abuse, bias, safety, privacy, and surveillance are all real concerns. They need serious policy, accountability, and protection. But if that is where the conversation ends, we miss the much bigger opportunity.

Women should not only be protected from AI. Women need pathways to participate in AI, lead with AI, build with AI, question AI, govern AI, and benefit economically from the changes AI is bringing to work.

This means we cannot treat AI readiness as an individual responsibility. We cannot simply tell women to be more confident, more curious, or more willing to try–the “Girlbossification” of AI, if you will. That kind of “fix the women” approach ignores the second-generation biases that show up in everyday workplace systems: who gets access to informal learning, who is seen as naturally technical, who gets sponsored into innovation work, who is penalized for taking risks, and who is expected to prove competence before being trusted with new tools.

It also ignores the reality of stereotype threat. If women already know they may be judged as less technical, less innovative, or less “naturally suited” to AI, experimenting publicly can carry more risk. A mistake with a new tool may not be read as a normal part of learning. It may be treated as confirmation of a stereotype. That changes who feels safe to try, who feels safe to ask questions, and who gets to build confidence over time.

That is part of why we wrote the op-ed. We wanted to put a clear point of view into the world at a moment when a lot of people are talking about AI strategy, but fewer are talking about what it actually takes for everyday workers, entrepreneurs, marketers, founders, and teams to use AI with confidence.

We wrote it because the next phase of AI adoption cannot be left to the loudest people in the room. It needs practical education, trusted support, responsible policy, and clear pathways for people who have historically been left out of technical change.

And… very honestly… we wrote it because this is the work Growclass is built for ✨

AI adoption needs a sandbox

So what the heck does an op-ed about national AI strategy have to do with a blog post about writing like a human? A whole heck of a lot more than you may initially think.

Because underneath both conversations is the same question: are people being given enough space, support, and trust to actually try?

That is a big part of what we were getting at in the op-ed. If AI adoption is treated like something that happens because a strategy exists, or because a tool is available, or because someone drops a “top 10 prompts” PDF into a shared drive and calls it training, we are setting people up to fail. Real adoption takes experimentation and practice and room to try something. You need to give yourself and your team permission and agency to get a weird result, ask a question, adjust, try again, and slowly build confidence.

That is especially important when we are talking about women and underrepresented people in AI. The issue is not a lack of interest or ability. It is often a lack of access, a lack of learning opportunities, and a very reasonable concern that if you try something imperfectly, you may be judged more harshly for it. That is why the op-ed argues that Canada’s AI strategy cannot stop at risk mitigation. If AI is really going to be “for all,” people need pathways to participate, lead, build, and benefit from it.

And honestly, you can see the same dynamic show up in how organizations are using AI to write.

This is how we get AI slop. Not because the tools are inherently bad, and not because people are lazy. We get it because people are told, “Put this prompt into ChatGPT to get this outcome,” and then they do exactly that. They do not change anything. They do not question whether it sounds like them. They do not have the confidence, time, training, or permission to experiment with the output. So everyone gets the same result over and over and over again.

Suddenly, Stephen sounds like Steven, who also sounds like Steve, who somehow also sounds exactly like Stephan. It is all the same thing in a different font.

That is not so much a writing problem as a culture problem.

If your team does not feel like they have space to try, they will cling to the safest-looking output. If your employees are not taught how to use AI as a thinking partner, they will use it like a vending machine. If your community members are not invited to experiment, question, remix, and bring their own judgment to the process, they will assume the first answer is the “right” answer.

And then we all sit around wondering why everything sounds the same.

I know I am saying this from a place of real privilege. I work at a company run by women who understand the creative process, who teach this stuff, and who genuinely believe that trying, experimenting, and making something a little weird before it gets good is part of the work. Not every team has that kind of culture yet.

But I think every organization needs it.

Because a tool can help you draft an email and get over that executive function stall. It can help you summarize a report. It can help you turn one long piece of writing into five different formats for five different audiences. That is genuinely useful, especially for small teams and busy people trying to do more with less.

But the tool cannot know what you owe your audience.

It cannot know that your funders need evidence, your members need reassurance, your sponsors need alignment, your founders need specificity, your students need clarity, and your community needs to feel like they are being invited into something instead of being marketed at. It cannot know which audience needs a little more context, which audience is already tired of the same talking points, and which audience will immediately clock that the “warm and approachable” paragraph was run through a kindness simulation.

That is the real writing work.

And it is also the real adoption work.

It is giving people enough support to try, enough context to make good decisions, enough room to fail safely, and enough confidence to bring their own judgment back into the process. Because writing like a human in the age of AI is not about rejecting the tool. It is about creating the conditions where people know what to do with it.

How we practice this at Growclass

At Growclass, we teach AI as a tool for strategy, systems, and support. It can help with the blank page, the messy research pile, the repetitive formatting, and the “how do I turn this into three versions for three audiences” spiral.

That is useful. But easier is not the same as better, and polished is not the same as personal.

Our AI Marketing & Strategy Certification is built around that belief. We help people understand how AI works, where it helps, where it falls short, and how to use it in ways that align with their brand, values, and real business needs.

We also try to practice this as a team. Every week, we have a meeting hold in our calendars called The Crab Lab, where we talk about AI, share what we have been trying, teach each other things, ask questions, and experiment. Sometimes it is a proper show-and-tell. Sometimes it is a place to bring the thing that did not work and figure out why. And honestly, sometimes it is just a chance to catch up and save ourselves from sending that one extra Slack clarifying message.

But the hold is there every week. Whoever can show up, shows up.

And that is the whole point.

Because the goal is not more generic content. The goal is more room for judgment, customer insight, creativity, experimentation, and the human decision-making that makes the work worth reading.

AI can help you think and organize. It cannot care about your audience for you.

A tiny partner toolkit for writing like a human again

We are working on a small resource for partners around this, because apparently we cannot help ourselves. We see a useful idea and immediately want to turn it into a toolkit. We just love to share the love, OK?!?!

For now, here is the working version. Before you publish something AI helped you write, run it through these five questions. I’ve used “AI adoption” as the example, but obviously insert your own problem worth solving!

1. Who is the actual human reading this?

Get more specific than “partners,” “members,” or “our audience.” Picture the person on the other side.

Is it a founder trying to make payroll? A marketing manager who has been told to “use AI” but has received no training? A funder scanning for outcomes? A sponsor deciding whether your community aligns with their brand? A student wondering if they belong in the room?

The more clearly you can see the person, the easier it is to write something useful for them.

2. What do they already know?

Good writing meets people where they are. That means knowing what context they already have, what they do not need explained again, and what might still be fuzzy.

This is where AI often gives you the average explanation. Your job is to make it the right explanation.

3. What are they worried about but not saying out loud?

When we think about AI adoption and AI education, for example, we try to ask what is already happening before we jump to solutions. What have people tried? Where are they getting stuck? What feels risky, confusing, or surprisingly useful? What would make trying feel safer?

Very The Mom Test coded (iykyk): don’t ask if people “care about AI literacy.” Ask about their real behaviour. The quiet worries are usually underneath: Am I behind? Am I allowed to use this? What if I use it wrong? What if everyone can tell AI helped me?

Those worries shape whether people try, learn, and keep going. Good writing names the real worry in the room.

4. What sentence could only we write?

This is my favourite test.

Read the draft and ask: could any organization have published this?

If yes, add more of you. Add the specific example, the tiny opinion, the lived detail, the line that sounds like your team, or the point of view that made you want to write the thing in the first place.

That is usually where the good stuff is hiding.

5. What decisions still belong to us?

AI can help with structure, phrasing, title options, summaries, and blind spots. Great. Use it.

But your ethics, promises, audience priorities, point of view, and relationship with the reader need to stay in human hands.

That is the difference between using AI to support your writing and letting AI sand off everything that made it worth reading.

The one-week “no AI writing” challenge

To get started, try going one week without using AI to write your actual words.

Use it for research, summaries, outlines, pressure testing, or clearing tasks you avoid. But write the first draft yourself. Or at least the first paragraph.

The goal is to notice what your own voice does before the internet’s average voice gets involved.

And yes, I’m doing it too. Old school. Writing my own emails alllllll by myself ​​😅🤪

What’s next

We are pulling together the Tiny Toolkit for Writing Like a Human Again that you can share with your team, members, community, or anyone else trying to use AI without sounding like the internet swallowed a corporate brochure.

You can also read the full Hill Times op-ed here.

And if your audience is trying to make sense of how AI is changing search, content, visibility, and discovery, you can share our upcoming Growclass event: Extra Credit: AEO and Search Visibility! 

Wednesday, Jul 22

9:30 AM - 12:00 AM EST

Because yes, AI is changing how people find answers. But the answer still needs to sound like it came from someone who understands the human on the other side.

Until next time… waving,
Talitha

‍