I have used AI a lot in my business this year, and it is both fast and a dangerous. I woke up to a study that MIT and Stanford conducted on something called delusional spiraling, and it made me wonder if all the work I’ve done has been for nothing. Now typing out that sentence I think I naturally have a layer that protects me because in some areas tools like Rank Math (a website SEO plugin) and the AI chatbot agree.
I have always used the adage trust but verify and in some ways I have kept to that, but in others I have swerved off the path a bit. As a former English major and someone that likes writing I was reluctant to not write my own blogs for my business, but I am more of a personal essayist than an SEO writer. I could never stay in rhythm when writing business blogs.
It made sense to turn the task over and I have had tight control over it, but I have loosened that control a bit. Before I was doing paragraph by paragraph read throughs and editing. Now I am doing more red teaming and letting the AI edit itself. I look more for weak language, couching statements, and common AI patterns than the deeper truth of the matter, and that could be a problem.
Now I have recognized it and the question is how do I handle it. The answer is to really examine the prompts used for writing. The AI needs to avoid statements not grounded in evidence except where it is a matter of opinion and my marketing material demands a particular opinion. For example we have a post coming up on Thursday about how to choose a pet sitter or dog walker, and I want our team approach highlighted as a strength, for obvious reasons.
In this day and age it is becoming even more important to demand statements be backed up by evidence. I fear that social media and other recent developments have primed people’s mental pump to be more susceptible to statements not backed by evidence. The point is to always ask for evidence unless something is a statement of opinion and then it’s just that, an opinion.
David Huzzard Writes 