[Brand](https://sustainablemarketing.com.au/blog/brand) [Digital](https://sustainablemarketing.com.au/blog/digital) [Marketing](https://sustainablemarketing.com.au/blog/marketing) [Public Relations](https://sustainablemarketing.com.au/blog/public-relations) # AI Slop Isn't a Technology Problem. This Week Proved It Twice. Posted 05 Aug 2026 by Maria Anderson ## AI Slop Isn't a Technology Problem. This Week Proved It Twice. **Key takeaways** - AI doesn't replace marketing expertise. It frees up time for the judgement and creativity that still need a human. - AI slop happens when people brief AI badly, or don't brief it at all, not because the tool is bad. - Better AI output doesn't come from a cleverer prompt. It comes from an AI framework, an operating system of sorts, built from a clear role, your knowledge and your guardrails. - A person with real marketing expertise using AI will always outperform someone without it, because they know when to dig deeper. - The process is the same with or without AI. What changes is the speed, not who's accountable for the result. This week I presented to Metro South Health, part of Queensland Health, on how AI is already changing marketing, communications and media. The message I kept coming back to was simple. AI isn't there to replace professional judgement or creativity. Those stay critical. What AI does is free up time, so there's more of it left for the work that actually needs a human. I believe that. I've also just watched it prove itself twice in the same week, in two very different ways. ### A knowledge kit, not a shortcut For the Metro South Health session, I wanted to show what good AI use actually looks like, not just talk about it. So I built a dummy knowledge kit for a press release scenario. I researched it through Perplexity, checking sources as I went, then briefed our own PR and Media Specialist tools, a custom GPT and a matching Claude skill, both built for Sustainable Marketing Services, to draft from it. I did it that way deliberately. A lot of government organisations don't yet have access to tools like ChatGPT or Claude. Some are about to gain access to Copilot instead. I wanted the demonstration to reflect the tools their teams will actually be using, not the tools I use. That detail turned out to matter more than I expected. ### Where AI slop actually comes from On the Monday before the presentation, a government client of mine in Canberra called. Their team will soon have access to Copilot, and they wanted to know if we could run some training. I asked how they were using AI at the moment. They open a chat, type something like "write an email to this client", and paste whatever comes back straight into their outbound email. No brief. No knowledge documents. No organisational framework at all. Sometimes the AI hands back a two thousand word email for a task that needed three sentences, and it goes out anyway. I want to be clear that this isn't a criticism of them. It's the same pattern I see in small businesses every week, not just government teams waiting on Copilot. A business owner opens ChatGPT before a proposal is due, types a rushed request, and sends back whatever it hands over because they're busy and it reads fine at a glance. Most people, in government and in small business, simply haven't been shown how to get a good outcome from AI. This is where AI slop comes from. Not from a bad tool, and not from any one sector. From no brief at all. It's also why so many people are unhappy with their AI results. The output sounds generic, and it sounds like everyone else's, because it's been briefed the same generic way everyone else is briefing it, which is to say, not briefed at all. Forbes ran a piece on exactly this earlier in 2026, arguing that one properly built knowledge document is what actually stops small businesses drowning in AI slop. I think that's right. ### What five hours instead of fifteen actually took The same week, I delivered a capability document to a client. Normally, the first draft of a document like that takes me around fifteen hours to write. This time, the first draft took five. The client reviewed it, asked for two minor changes, and approved it. They were genuinely happy with it. As a marketing expert, I was proud to present that draft. It was good work, not just fast work. That result is real, but it isn't the whole story, and I don't want to leave it sounding like a magic trick. Last year, plenty of AI tools were sold on the promise that you could do a month of social media in fifteen minutes. We tested some of that software ourselves. It doesn't work like that, and AI slop delivered faster is still AI slop. You still have to go through the process, and you still have to invest time to save time. The five hours only happened because of work I'd already done long before I opened a chat window. I'd built that client a documented brain of their business, a chief marketing officer AI trained on their strategy, a brand style guide, and a writing tool built specifically for their industry. The speed came after the investment, not instead of it. ### Why one assistant can't do everything Somewhere in building that system, I noticed a mistake I now see constantly. People try to make one AI do everything. Think about how that would work with a team member. You wouldn't ask the same person to be your chief financial officer one minute, your chief marketing officer the next, then your business strategist, then your operations lead, then hand them the writing on top of all of it. No one could do that role well, human or AI. Over time, that's why we built a small set of AI specialists for our own business, each with one clear job and clear boundaries, sitting under a strategist that checks everything stays aligned with how we actually work. Clear roles. Clear boundaries. Better decisions. Maria Anderson, Founder and Managing Director of Sustainable Marketing Services ### Better AI doesn't come from a better prompt What actually drives a good result isn't a cleverer prompt. It's having an AI framework, call it an operating system, built from a few connected parts that you feed into every task. At the centre is an intelligence layer, business intelligence or strategic intelligence, whatever you want to call it, that captures how your business actually thinks and operates. Around that sits your branding documentation, your audience and ideal client insight, examples of your best work, your standard operating procedures, clear roles for what each AI tool is there to do, and guardrails for what it must never do. You write that once and reuse it every time. The businesses that take AI seriously are the ones investing in that documentation now. Brand strategy, brand positioning, your ideal client profile, a marketing plan and writing guide, a brand style guide, your marketing systems and offers, your standard operating procedures. None of that is busywork. It's what gives AI something real to work from, and it's increasingly what separates a business getting a genuine edge from AI from one that's just generating average content faster than it used to. Once that foundation is solid, the next step is looking at what can be safely semi-automated or automated, but that's a conversation for further down the track, not the starting point. And it means a human who stays involved through the whole process, not just at the end. It's not a sign off gate you check on your way out the door. It's someone driving the process from brief through to final check, the same way you would if AI wasn't involved at all. Guardrails matter here too. AI should never invent quotes, facts or statistics. It should never be handed sensitive information. It should never publish anything before a human has reviewed it. Those aren't limits on what AI can do. They're what make the outcome trustworthy enough to use. ### Expertise still decides who wins Here's the part I think gets missed most often. Put two people in front of the same AI tool, both writing something for marketing. One is a marketing expert. One isn't. The marketing expert wins, every time. Not because their prompt is better. Because they know when the draft needs to go deeper, when the tone is wrong for the audience, when a claim needs proof behind it, and when something that reads perfectly well is still the wrong thing to say. Without that expertise, you can't tell the difference between a good draft and confident AI slop. It's something I talked through with a client this week. I've built AI tools for their business, and without some training on how to brief and use them well, they won't get the same outcome from those tools that I would. That's really the whole point of AI². Actual intelligence, your expertise, plus artificial intelligence, the tool, delivers a magnified outcome. Not just a faster one. ### It's the same task, just faster Whether you write something yourself or write it with AI, the process doesn't change. Understand your goal. Understand your audience. Write a proper brief. Gather your knowledge and research. Draft it. Check the facts. Refine it. Get sign off. AI can help with the draft. At a more advanced, agentic level, it can help with the research too. But the steps are the same steps, and the person accountable for what goes out the door hasn't changed either. The difference is speed, not judgement. ### Know which tool to use for what Part of getting a good result is knowing that AI tools aren't interchangeable. They're moving fast, and they each have different strengths. At the moment, I lean on Perplexity and Gemini for research, with Perplexity ahead for most of what I need. For analysing detailed material, I've found ChatGPT does the strongest job. For writing, once it has a clear brief and good knowledge documents behind it, Claude is hard to beat. Matching the tool to the task matters as much as briefing it well. ### Where to actually start If all of this feels like a lot, start small. Document what you know before you worry about which tool to buy. One low risk task. One shared document your whole team can reuse. Give the AI a clear role and ask it to question you before it drafts anything, rather than accepting the first thing it hands back. If a chat starts drifting or getting less accurate the longer it runs, that's not a flaw you need to fix. It just means the conversation has outgrown what the AI can hold onto. Open a new one for the next task. Better AI starts with better knowledge. Document what you know, treat AI like a smart, fast colleague, give it clear goals, check its facts, and keep the sign off with a human. Start with one task and one document, and build from there. ### Let's talk about what that could look like for you If you'd like to think through what a properly briefed AI approach could look like in your marketing, we're happy to have that conversation. For some clients that starts with a single documented brief. For others it becomes part of a broader marketing growth plan. Either way, it starts with a conversation, not a sales pitch. If that's useful, book a discovery call and we'll talk it through. AI slop is generic, low value content produced when AI is used without a clear brief, brand knowledge or guardrails. It happens because the AI has nothing specific to work from, so it defaults to the same generic patterns every other unbriefed user gets. No. AI can take over repetitive tasks, but it can't replace the judgement that tells you when a draft needs more depth, when a claim needs proof, or when the tone is wrong. Without marketing expertise behind it, AI use tends to produce more AI slop, not less. Not if it's briefed properly. In fact, a properly briefed AI can improve the consistency of your brand voice, because it works from the same documented voice, messages and examples every time. Without that brief, it defaults to a generic voice that could belong to any business. Using AI is opening a chat and typing a request. An AI framework is a documented system of knowledge, guardrails and clear roles that AI works from every time, so results are consistent rather than a gamble. Start with one low risk task and one shared knowledge document. Give AI a clear role, ask it to question you before drafting, and keep a human checking and driving the process throughout. No. The biggest factor isn't the tool, it's whether you've documented your brand, audience and guardrails before you start. A well briefed standard chat will outperform an expensive tool used without a brief. ## About the Author Maria Anderson is the Founder and Managing Director of Sustainable Marketing Services, where she helps growth-focused businesses make sustainable growth achievable through clear strategy, structured marketing systems, and practical leadership. With more than 25 years' experience in marketing, Maria specialises in turning fragmented marketing into a connected, measurable system that gives business owners confidence in where to focus, what to prioritise, and how to grow sustainably. A Certified Practising Marketer (Australian Marketing Institute), Maria holds a Bachelor of Business (Marketing and Public Relations) together with Diplomas in Sustainability and Financial Services. She combines strategic thinking with practical implementation, helping businesses simplify marketing, strengthen their positioning, and achieve measurable long-term growth. Maria is also the creator of the Sustainable Growth System, a strategy-led approach that connects strategy, leadership, execution, and optimisation into one practical marketing framework. She believes great marketing should never feel overwhelming. It should be clear, manageable, measurable, and achievable.