Spotting AI Part 2: Intros and the Obvious

I asked ChatGPT to write a roofing blog. The prompt I used for this example was “Can you write me a blog on the importance of a roof inspection after a storm?” Here’s the intro of the blog it generated.

The Importance of a Roof Inspection After a Storm

When a storm rolls through, it leaves a trail of debris, water damage, and structural wear in its wake. One of the most vulnerable parts of your home during such events is the roof. As the first line of defense against the elements, your roof bears the brunt of wind, rain, hail, and even lightning. That’s why scheduling a roof inspection after a storm is crucial. Here are some key reasons why this proactive step is so important.

The “…first line of defense…” line is a dead giveaway that this is AI. A quick Google Search will show you why.

I tried to cycle through SERPs to figure out how deep an exact match of that line went, and I gave up by the time I reached the 30th page. Most roofing websites have, at some point (likely more than once), described roofs as “your home’s first line of defense” in their copy.

It’s the kind of shallow, obvious marketing platitude generative AI loves to write because it has trawled the phrase thousands of times before in published roofing content.

The last line of the first paragraph is another very obvious AI indicator. Every AI blog will end the intro paragraph with some variation of “Here are some key reasons why…” or “This blog will explain…” or “Understanding x is key to determine…”

Just to further illustrate this point, here are six blog prompts and the last sentence of the intro section generated by ChatGPT:

• How pharmaceutical drug prices are determined: “This blog will explore the key elements that influence how pharmaceutical drug prices are determined.”
  
  
• What's the difference between type 1 and type 2 diabetes: “This blog will explore the key differences between Type 1 and Type 2 diabetes.”
  
  
• How to tell if I need to replace my tires: “Here are key indicators that it's time to get new tires.”
  
  
• What's the difference between slip and fall and premises liability?: “While both deal with injuries occurring on someone else's property, understanding the distinctions between the two can clarify the legal processes involved in each case.”
  
  
• What to look for in a wedding venue: “Here are key factors to consider when selecting your wedding venue.”
  
  
• What's included in HVAC maintenance: “Here’s a comprehensive guide to what’s typically included in HVAC maintenance.”

Obvious Sections and Topics

Another AI giveaway is the basic regurgitation of superficial, valueless talking points. The sections for the storm roof inspection blog included:

• Detect hidden damage
• Prevent costly repairs
• Ensure insurance claims are handled properly
• Maintain home value
• Safeguard against future storms
• Protect your family’s health

Generative AI home services content is particularly obvious when the topics are formulaic. In AI’s defense, the topic of storm damage roof inspections is painfully basic. It is possible to get better copy out of ChatGPT by using topics that haven’t been covered a thousand times in previously published website and blog copy.

For example, the “How pharmaceutical drug prices are determined” blog wasn’t bad because it’s not a topic commonly written about in blog form. It’s also a broad enough topic that the AI did not get overly repetitive in the 500 or so words it gave me.

Advice No One With an Ounce of Real-World Experience Would Follow

People who rely on AI for content generation sometimes fail to remember that the generative chatbot has no actual real-world experience. It assumes what businesses publish on their website is reflective of reality and homeownership best practices. 

A good illustration of this phenomenon is blog or website FAQ content that answers the question, “How often should I have my roof inspected?” AI will invariably say once or twice a year, typically once in spring and once in fall.

Perplexity.ai will give you similar results, “At least once a year…” and “Twice a year is even better…”

How many homeowners actually have their roof inspected annually? Or even every other year?

The vast majority of homeowners don’t have their roof inspected until they have a leak, or they notice damage when looking up at their roof from the street or their driveway. In the real world, failing to have your roof inspected annually never results in catastrophic roof failure.

All ChatGPT knows about roofing is what it has learned from trawling roofer content available online. All that content is pushing the same marketing narrative. You shouldn’t expect generative AI to know better.

Content is far more likely to be written by a human if it instead recommends a more realistic roof inspection schedule, like:

“You probably don’t need an inspection for at least five years after you replace your roof, unless your home has been impacted by a severe storm or you notice missing shingles after high winds. If you do see roof damage within five years of roof replacement, despite no severe weather events in that time, chances are your roofer messed up and you should call them to fix it. Once an asphalt shingle roof is 10 years old or older, you should consider scheduling inspections once every two years…”

Consumers are not dumb. One of the core problems with AI-generated content is it inadvertently speaks to readers as if they are.

I always discourage my writers from leading into body sections with something like “Here’s a comprehensive guide…” or “Understanding the distinctions between…” or “This blog will explore…” because it’s a fundamentally condescending way to write. People don’t need to be spoon-fed the blog – and it’s just a needlessly expositional waste of words.

The intro should set up the topic while organically leading into the first subheading of the body. Readers can infer from a good H1 and intro paragraph what the body will cover. Unfortunately, AI has been trained on lazy, patronizing copy written to meet word count targets and trick people into wasting money on products and services they don’t need. Most consumers can see through that. 

This approach is also not great for the reputation of marketing clients. It can come off as transparently manipulative, and it’s best not to assume readers and potential customers are that gullible.

This flawed approach to content formulation is not the AI’s fault. It’s the fault of content writers, editors and lazy SEO specialists who can’t balance the needs of campaigns with originality and an honest perspective.

Get Help With Your Content Marketing Strategy

REV77 is not anti-AI. To the contrary, we see the value it offers when used as a tool to enhance efficiency and the depth of research. Unfortunately, many digital marketing providers are defaulting to AI as a replacement for human writers. Companies who begin relying solely on generative AI for content may start to run into problems, not just in terms of repetitive, low-value content but also in the way it presents its clients to their prospective customers.

If you’d like to benefit from AI without sacrificing the quality or relevance of your website content, REV77 can help. Contact us for a free digital marketing audit.