AI Policies

AI is currently being used as an umbrella marketing term. These AI usage policies reflect a complex array of tools, some of which pre-date the current AI trend. This is not yet a comprehensive list as the technology is changing rapidly. I will be using the accurate term LLM (Large Language Model) to describe products like ChatGPT, Gemini, Deepseek, Claude, etc.

Generative AI – LLMS and image generators

No. Absolutely not. I have been transparent about the fact that I tested early versions of ChatGPT and Claude. After reading further into the copyright, intellectual property, theft, cognitive and ecological issues, I determined that it was a tool that has no place in my business or work. This step helps to protect both my work and yours. Additionally, I have never used AI image generators and will not.

Transcription and notetaking services

To date, I use the transcript feature in Google Meet only with permission from clients. When using Zoom, I use the internal notetaker only with permission or upon request. I also offer video recordings also upon request. I do not use any external notetakers as they have proven to have security and accuracy issues.

OCRs and other text detectors in images

I’ll be honest, it is weird to even add this as an AI category, but here we are. OCRs (Optical character recognition) are machine learning tools that can scan and pull text from images and PDFs. The technology has been around since the 1970s (fun fact: I first read about them in Congo by John Grisham). I do use them to search text in images and PDFs and I also use them when available to copy text in an image and add it as alt text.

AI automations

I used Claude/ChatGPT/Gemini to write my blogs and product descriptions, can you edit them for me to sound more human?

Writing is work. I get that it is real work because I do it. There is an esoteric quality to writing. Putting your swirling thoughts into concrete words certainly feels like a struggle. But it is also how you develop your point of view (POV). How, why and when you choose certain words and certain thoughts and why you think they convey the meaning you want them to is part of the process. And yes, this is true even when it’s two to three sentences describing your work. In those cases, POV matters as much as grammar. When an LLM writes for you, that POV evaporates. Once that is gone, there is nothing to edit, because I cannot ask you what you were thinking when you wrote it. All of this is a long way of saying that my current policy is not to edit any output from any of the LLMs.

I don’t currently use any of the AI automation features in IFTTT, Zapier or Calendly. I do use webhook/API automations to do things like connect my eCourse platform to my email marketing platform after the opt in. I use some of Zapier’s original automations and existing automations in Calendly to send calendar reminders. I have often been asked about predictive (“AI”) tools for inventory. To that I ask, have you done a physical inventory yet? Do you understand what your inventory numbers are telling you? If not, a machine learning analysis can’t provide accurate or useful intel. To date, I have been unimpressed with the stability, functionality, and ease of the AI automation features.

Spellcheck and grammar

I stopped using Grammarly because their AI features made their product worse. I currently use whichever spellcheck is native to the browser or software that I am using. I also read through my text because spellcheck isn’t always accurate in catching homonyms. When I need a grammar check, I use the paid version of the Hemmingway app that is downloaded to my computer and is free of any AI add ons.

Chatbots and AI assistants

Good god no. Across the board as I have tested these tools not a single one has been useful. Gemini was a hot mess that disrupted my work by showing me pop up suggestions every few minutes. I turned it off after—you guessed it!—*I had to have a lengthy chat with a chatbot that went nowhere until I got a human customer service agent. I also turned off Studio because it is designed to scrape my emails, and that would have exposed too much private data – including yours. If you, like me, are old enough to remember Clippy, this is like if Clippy couldn’t be turned off and was also stealing your data. And if you remember Clippy, get your mammogram and your colonoscopy.

*Right about now you are wondering how an em dash got in here when I don’t use generative AI and that’s because I’ve been using them for 20+ years. AI can pry the em dash out of my cold dead hands.