By Natalie Khoo
For years, AI (or, more accurately, machine learning) has been transforming the marketing and communications industry. Tools like Grammarly, SEMrush, and predictive text have become everyday essentials for marketers.
Until recently, there wasn’t much debate about whether to use these tools… Everyone just did. The conversation was more about which tools worked best.
But with the rapid rise of generative AI tools, the landscape has changed. We’re no longer just talking about convenience or efficiency. Now, the real question is: how do we use AI responsibly?
At Avion, we’re passionate about upskilling in emerging technologies in a practical and human-centred way. Recently, we’ve been exploring:
What does it mean to use AI responsibly?
How can we use it while preserving the quality of our work?
How can agencies build trust with clients when using AI for certain tasks?
What to include in your AI usage policy
We believe that ethical AI usage isn’t just about compliance. It’s about setting a standard that helps everyone do their best work – while ensuring alignment of expectations across teams. That’s why we have an internal AI usage policy that all employees are required to sign off on.
Avion is not a provider of legal advice. However, here are some overarching principles that we’ve included in our AI policy that might inspire your next steps.
Responsibility: putting confidentiality and trust first
We have a commitment to protecting client data and never enter it into public AI tools. AI is only used when necessary and in alignment with a client’s policies.
We also believe in transparency, and this means being honest about how AI may be used in client projects. (If we do use AI as part of our work, every output is reviewed and refined by our experienced copywriters – more on this below.)
Implementation: human-in-the-loop and ethical AI practices
AI is a powerful tool, but for us it never replaces human expertise. We follow a human-in-the-loop (HITL) approach, where AI plays a supporting role.
In short, this means we only use AI to assist with tasks that may help the content creation process. All final outputs and strategic thinking remain firmly in human hands. Our content writers always remain in control.
If AI is used to facilitate our work, here are our team’s most common applications.
Using generative AI for early-stage content planning. For example:
Brainstorming topics and potential outlines to work with
Checking for similar content already published by competitors
Identifying gaps and opportunities in content briefs.
Using generative AI to interpret complex ideas. For example:
Summarising key points from publicly available reports
Finding simple ways to explain unknown or complicated concepts
Searching for answers quickly within lengthy documents (i.e. terms and conditions).
Using generative AI to conduct valid research. For example:
Fact-checking statements and claims
Finding up-to-date research from credible sources
Discovering what the industry is saying about a particular topic.
Using generative AI to refine human-written content. For example:
Exploring different options for tone of voice
Asking for suggestions on how to improve a readability score
Converting Australian to US spelling and grammar (and vice versa).
Our team is also aware that AI is constantly evolving. We continually look out for new tools and trends, adapting as we go, and discuss these as a team (in a Slack channel uniquely created for AI updates, as well as in monthly team forums). This not only helps us individually as content professionals, but it ensures we have time to regularly reflect on our AI use, so it continues to meet the needs of the agency and our clients.
Commitment: people before technology
Our ethos is simple: people come first. ChatGPT will never have the same ability to “feel” the way we do. Our content strategists and copywriters do more than just write for tone of voice – we get to know the audience, we do our best to empathise with what they’re going through, we try to understand what they need, and we follow their experience across a website or multiple channels.
Avion’s AI policy development process
So how did we get here? We started as we always do, with collaboration.
Our first step involved a workshop, in which team members shared how they were already using AI, alongside ethical considerations. Together, we explored what would work best for our clients, and how we could find the right balance between what humans and AI do best.
Then we moved onto drafting the policy. We consolidated the team’s input before presenting it for further discussion and feedback and making it official.
Team workshop: discuss current AI practices and ethical considerations
↓
Policy draft: consolidate insights and draft the policy
↓
Team feedback: share the draft for review and refine collaboratively
↓
Finalisation: publish the policy and socialise it with the team
Practical tips for creating your own AI policy for work
If you’re looking to implement an AI policy for your team, here are a few activities we’ve found helpful:
Define your core principles: Identify your ethical stance on data protection and transparency, and factor that into your thinking as you decide AI’s role in your business.
Set boundaries: Outline where and how AI can be used, emphasising HITL workflows, and avoiding sensitive data input into public tools.
Communicate clearly: Ensure your team and clients understand your policy, why it matters, and how it’s implemented.
Collaborate and evolve: Engage your team in discussions, gather feedback, and update the policy as technology and business needs change. Your policy must evolve just as the technology does.
Image credits: Illustration by pch.vector on Freepik. Custom GIFs by Digital Pratik and blinkdotla on Tenor.