Skip to content
(e.g.) creative content (e.g.) creative content

Generative AI Policy

Use of AI in original work or work for hire

I pledge to use no generative AI in either my original content or my clients’ content barring the following exceptions:

  1. Spellcheck tools such as the built-in Word/Copilot spellcheck or licensed services such as ProWritingAid
  2. Testing and/or generating screenshots of an AI capability (or non-capability) for the purposes of explaining an AI-related concept or step in an article
  3. Transcription, only in situations where all parties privy to a call are aware of the recording and need for transcription. Purpose-built and paid, licensed tools such as the Riverside video meeting recording/transcription capability will be used for this purpose; call audio and video will not be uploaded to general-purpose free chat service. Offline, local transcription models will be used wherever possible.
  4. Translation between languages; not for publication or external consumption but for internal analysis and learning purposes.

This means that generative AI will not be used for:

Use of AI in collaborative projects

Furthermore, I do not knowingly accept:

from clients seeking to engage my services in a content project, however I recognize that this is, to some extent, out of the scope of my control. Therefore, this policy primarily concerns generative AI use and deployment within my direct scope of control. 

Use of non-generative machine learning technologies

Non-generative machine learning components contained in tools including but not limited to Semrush, Ahrefs, Riverside, Clearscope, and Microsoft Office will be used for:

Any images and video contained in content I produce for my business or clients will be obtained via open-source, creative-commons attributed libraries or licensed stock services and will not, in good faith, be selected if suspected to involve AI assets. When I use tools like Microsoft Word and Canva, both of which contain generative technologies, I do not generate content or assets. Instead, I produce my work manually as I did before the introduction of generative technologies in these tools.

This policy is not legally binding but a statement of my intent and plan of action as I move forward in an AI-heavy marketing world. It is subject to change over time and I will update my policy accordingly as tools and techniques evolve. I commit, however, to continuing to not knowingly use generative AI technologies as outlined above. 

Why am I creating this policy?

  1. Generative AI is rapidly entering the majority of marketing technology tools available today, and is heavily relied upon by service providers and agencies around the world. When someone isn’t given transparency into how the person they contract to support their business uses generative AI tools, it can create matters of data risk, opportunities for plagiarism, and expose private details to public audiences. My clients should not need to concern themselves with the security of the information they share with me, whether or not we’ve signed an MNDA.
  2. While non-generative AI and machine learning applications have existed for decades and have many useful real-world applications, the rollout of generative AI (starting in 2022 with the public launch of ChatGPT for purposes of this policy) has not proven to be similarly useful, with only 26% of people finding that it actually helps them write content significantly faster and 80% of enterprises reporting no measurable impact on EBIT. This is creating a world in which companies are over-leveraging their investments into AI without seeing the necessary or promised financial return, ultimately creating situations where losses or shareholder dissatisfaction purportedly necessitates job cuts. I find this morally distasteful and do not support it.
  3. The large language models (LLMs) that power AI tools are trained on copious amounts of pre-existing literature, audio, video, web content, other creative / copyrighted assets. This information has largely been gathered by AI companies without permission or recompense to the copyright holders. For example, if you wanted to store and read all of the books that Meta used to train its AI, without permission from copyright holders, you would need over 5,000 Kindle Paperwhite e-readers. This copyrighted, human-created work is then repurposed into generative AI outputs. I am ethically opposed to this mass theft of creative and professional work. 
  4. Generative AI only works with the presence of large physical data centers that create water, light, air, and noise pollution for the communities in which they are located. In the United States, state governments typically give large tax breaks to data center companies, but when these data centers are built they reduce collectible property tax (which funds public school districts), drive up the cost of residential electricity bills, and keep people awake at night through light and/or noise, the latter often from generators. I personally live half a mile (~805 meters) from a data center and can hear the noise of the generators when they’re fired up for testing or power support. I do not support the proliferation of these data centers, which typically employ under 100 people, in any community.
  5. The leaders of generative AI companies typically engage in deceptive practices such as claiming they are close to “AGI” when their companies actually maintain a separate definition of AGI—case in point being OpenAI repeatedly referring to “AGI” (originally a sci-fi-esque concept of machines achieving higher intelligence than humans) but not publicly clarifying that their internal definition of AGI refers not to intelligence capabilities but revenue earned through the use of ChatGPT models. I find this distasteful and prefer not to support companies engaging in these practices.

How am I putting this into practice?

I am not using generative AI as outlined above, with minor exceptions as noted (i.e. proofreading using a tool that is technically a generative AI product, but not generating new content with it, or generating an asset to use as an example in an article like my pieces on why AI content isn’t great for SEO).

Does my refusal to use generative AI cost my clients more?

No. Even when working with me on an hourly or retainer basis, my clients are not spending more money by paying me to not use generative AI. I have been doing my job for over a decade; I have tested and shown that I work faster without generative AI as my processes are well-defined, honed to precision, and proven through repeated use.