Answers

Should I use AI to write my content?

Yes, use AI to write your content, with one firm limit: it drafts, you decide. The businesses that get burned are the ones who paste raw model output onto a live page and call it finished, because that produces the same undifferentiated text every competitor is generating from the same handful of models.

So when someone asks should I use AI to write my content, the honest answer is a qualified yes. Use it for the mechanical work. Keep yourself on the parts that give a page a reason to exist: firsthand experience and a point of view. A model is fast at structure and first drafts. It is weak at the two things that actually make a page worth reading.

What Google actually rewards here

Google does not penalize content because a machine helped write it. Its guidance has been consistent: it rewards helpful, people-first content and suppresses thin, derivative pages regardless of how they were produced. The risk was never the tool. The risk is publishing something that says nothing a reader could not get from ten other pages.

That distinction kills the laziest use of AI. Ask a model to write a 1500 word article about a topic and you get a confident, fluent, generic page. It reads fine and earns no attention, because it carries no information gain. The pages that earn attention add something a model would not volunteer: a real process, a specific number you actually have, a decision framework, an opinion with a spine.

Information gain is the whole concept. It is the material on the page that nobody else covering the topic has bothered to include. AI, by design, averages what already exists, so on its own it cannot manufacture gain. You bring the gain. The model helps you type it up.

Where AI genuinely helps

Used as an assistant rather than an author, AI removes hours of mechanical work:

• Outlining: turning a messy brief into a logical heading order before you write a word.
• First drafts of mechanical sections: definitions, glossary entries, FAQ scaffolding.
• Reformatting: compressing a transcript or a long email into tight web copy.
• Research triage: summarizing what already exists so you can find the gap worth filling.
• Variations: rewriting one proven paragraph into several angles to test.
• Internal linking suggestions: proposing related pages to connect once the draft exists.

In every one of those, a person still supplies the judgment. AI handles the typing. You handle the truth.

Where AI quietly hurts you

• Invented specifics: models fabricate statistics, studies, and quotes with total confidence. Every number has to be checked against a source you can name.
• Flattened voice: unedited output converges on a bland, hedge-everything register that makes every brand sound identical.
• False authority: a model cannot describe your process, your customers, or your results, so it fills the gap with plausible filler.
• Scale without strategy: it is trivial to publish hundreds of pages no one needs, which only dilutes the handful that matter.
• Sameness at scale: when everyone prompts the same models, unedited pages converge, and your site starts to read like your competitors' sites.

How we treat AI in our own process

At Ninetyfifth Marketing we use AI the way a good editor uses a fast junior writer: to move quickly through the mechanical work so the human hours land on the parts that cannot be faked. Every page starts from real inputs about the business, gets drafted with assistance, then is edited line by line for accuracy and voice and fact-checked before it ships.

We also treat AI search and traditional search as one surface. The same honesty that keeps a page genuinely useful to a reader is what makes it safe for an AI answer engine to quote. If you want the longer version of how that fits into a build, see /services/seo/ and /services/ai-search-optimization/. The tool is fine. The discipline around the tool is the whole game.

When writing from scratch beats editing a draft

There is a point where AI stops saving time. For short, high-stakes pages where voice carries the whole message (a home page, a founder letter, a pitch), fixing a bland machine draft takes longer than writing from a blank page. The draft gives you something to react to, which feels productive, but you end up rewriting every line anyway. Reserve AI for the places where volume and structure dominate: answer pages, FAQs, definitions, documentation, and the long tail of questions your customers actually type. Match the tool to the job instead of forcing it onto everything.

A rule you can apply today

Before anything AI-assisted goes live, ask one question: what does this page contain that a reader could not get by typing the same prompt themselves? If the answer is nothing, it is not ready, no matter how clean the prose looks. If the answer is a real process, a verified fact, or a genuine opinion, then the fact that AI helped you write it faster is irrelevant.