Artificial intelligence is one of the most discussed topics in the business world. There’s a good reason for this: it’s genuinely useful. But it’s also easy to misuse, and it’s hugely misunderstood by a lot of the people who would benefit more from it were they using it correctly. If you’ve been wondering whether AI has a place in your marketing and what that might look like, this blog is for you.

Polly Want A Cracker?

Before getting into how AI applies to marketing specifically, it helps to understand what AI tools like ChatGPT, Claude, and Gemini are doing when you use them. A metaphor we use is that AI is very much like a clever parrot.

A parrot can recite an enormous number of things it’s been taught and sound like it understands what it’s saying. AI can do that, and in addition combine those teachings in new ways that can sound remarkably intelligent and original. But just like a parrot, it has no idea whether what it’s saying is true. It’s pattern-matching and predicting what comes next based on everything it’s been trained on, and it will deliver a wrong answer with the exact same confident tone as a correct one. This is called an AI “hallucination,” and it’s a fundamental characteristic of how these systems work.

Large language models (LLM), which is the technical category these systems belong to, are trained on enormous amounts of text they scrape from the internet, books, articles, and other sources. They learn statistical patterns in language, which word tends to follow which other word in which contexts, and they generate responses by predicting the most plausible continuation of what you’ve given them. They’re very good at sounding coherent and knowledgeable, but they’re less than reliable when the truth matters.

For marketing, this means it’s essential that anything AI drafts for you needs to be reviewed by a human before it reaches your audience. Incorrect content that goes out under your business name can undermine all four parts of Google’s EEAT framework at the same time, signaling to both search engines and real customers that your business lacks the experience, expertise, authority, and trustworthiness they’re looking for. A customer who catches an error on your website will think your business doesn’t know what it’s talking about.

So What Is AI Good For?

A clear framework for thinking about where AI works well is B-A-D: Brainstorm, Analyze, and Draft. These three functions are where these tools can deliver real value.

Brainstorming with AI is incredibly useful. If you’re stuck on what topic to write about, what angle to take on a campaign, or what to call a new business or service, AI can generate a quick list of options that you can work through and refine. In many ways it’s a thinking partner that never runs out of ideas – even if not all of them are good ones.

AI is particularly useful when you give it something concrete to analyze. You can paste in a competitor’s website copy and ask what messaging angles they’re emphasizing or paste in your own website and ask what a first-time visitor might find confusing. You can give it a set of customer reviews and ask it to identify common themes or ask it to compare a set of products and their features. AI processes text quickly and can find and clarify patterns that would take much longer to do on your own.

Drafting is where most people start, and this is where AI shines…with one important caveat. AI can produce a solid first draft of a social media post, a newsletter intro, a blog outline, or a product description faster than a human can. The caveat is that what it produces is indeed only a draft, and it needs a human edit to make it accurate, specific, and authentic before it goes anywhere near an audience.

Where AI Causes Problems

The problems with using AI in marketing tend to occur in a couple of specific areas.

Fabricated facts are the most dangerous. AI will sometimes invent statistics, misattribute quotes, name businesses that don’t exist, and cite sources that are not there, all while sounding completely credible and confident in its statements. We wrote about this in our AI content verification post, and we’ve seen it firsthand with our clients. If there’s a fact in an AI draft that you’re not completely sure of, verify it independently before believing it.

Confidential information is the other major problem. Whatever you type into a commercial AI tool is processed by its systems, and you should be careful about what you share. Customer data, proprietary business information, employee details, and anything you’d want to keep private should stay out of these conversations.

Let AI Write the First Draft, Not the Last One

One of the most common places we see problematic use of AI in marketing is using it unedited for website copy, and it’s a problem we want to address directly because the consequences can be longer lasting and more severe than just a bad social media post.

A social media post that misses the mark most likely disappears in less than a week. Website copy that was written by AI and never properly reviewed can sit on your website for a long time, continuously doing damage you’re not aware of.

There are three main problems with AI-written web content.

The first is that AI’s copy sounds professional but is generic; it says nothing specific about your business and what unique things you offer. AI has no idea what makes your company different from the others in your market that are offering the same service. It will write something that sounds like it should be on a website and hits all the main points, but it won’t tell a customer why they should choose you over your competitor up the street. Visitors can feel how vague the copy is, and they will leave. We’ve written more in depth about this aspect in our post on canned and AI-generated content.

The second problem is that AI is sometimes factually challenged. This is a big one, and easily fixable when one pays attention. We’ve seen AI-generated website copy talk about services a business doesn’t offer, quote prices that are wrong, and make claims about certifications or credentials the business just doesn’t have. AI is trying to fill in gaps in its understanding with what seems plausible based on similar businesses it sees online. But just because something seems plausible does not make it accurate, and when a customer is deciding based on what your website says that can be a real problem.

The third problem is that this content can severely affect SEO. Google’s EEAT framework evaluates whether content demonstrates real experience, expertise, and trustworthiness. Google recognizes and scores this kind of generic AI-written content poorly on all three of those markers. Such writing may look human at a glance, but search engines are getting better at identifying it, and businesses that invest in credible, human-touched content are the ones that continue to rank well.

AI can still play a big role in your website copy. Its ability to help draft, outline, and brainstorm what should go on a page can be an invaluable timesaver. But a human who knows and understands your business needs to be involved. Claims need to be verified, accuracy needs to be checked, and the final copy needs to sound like you.

Give Good, Get Good

The single biggest factor in how useful AI is for your marketing is how good your prompts are. AI can’t read minds, and if it gets a vague prompt it will produce vague writing and is more likely to try and fill in those gaps we just talked about. We use a framework called T.A.S.K. as a useful way to think about this.

Tell: what does AI need to know about your business, your audience, and the context of what you’re asking for?
Audience: define who this is for and what they care about.
Specific assignment: what exactly do you want it to produce?
Key constraints: what tone, length, and format, and what things to avoid.

The difference between “write me a Facebook post about our fall sale” and a prompt that includes your business name, what you sell, what the sale involves, who your customers are, and what action you want them to take is the difference between a generic post that will get no attention and one that will get people to visit your site and convert.