AI Use Disclosure.

I think all authors and creators should disclose what they have used AI for in their end product. This is not only important so that humans can make their own decisions about how reliable the materials are, but also because it prevents AI from training on AI-generated materials (or at least lets the training system know that it is training on AI-generated materials.

The videos are all me, but some of the graphics in them are AI-generated (as if you couldn’t tell already). I’ve used AI as a secondary way to check accuracy in some videos and articles. I’ve used AI to help me format things.

Some portions of the articles are written with the help of AI, but only in the form of “here is a transcript, please convert it into an article without adding any material I haven’t discussed except you should also include that there is one year to file a utility application after getting the provisional on file, no claims must be made in the provisional but it is best practices to include them, and examiners find figures helpful, so you might as well include them in your provisional.”

With really short videos, I normally write up additional material myself. Almost everything is put through AI to look for logic errors, typos, grammar errors, etc., and usually I ask AI to format it and make those kinds of fixes. I’m terrible at formatting things well, so I do have AI help me format more or less everything.

I encourage others to adopt an AI Use Disclosure. This is particularly important so that humans can come to their own decisions about how trustworthy and well researched an article or video is. For example, I’ll never ask AI “I want to do an article about how orbital period for planets are impacted by their rotational speed if at all, please research and write it for me” (my guess is no impact, but I have no idea). People who are looking to learn about something are easily able to ask an AI about it themselves, and I have no business passing AI work off as mine. Fixing my dense and complex writing style so it is easier to read? Sure thing, that’s a great AI use. Pretending I know things that I don’t? That’s never an OK use of AI in my book.

For some hypertechnical issues, where precision is crucial, I will use more AI resources than normal. I’ll also disclose it in the article. Some non-hypertechnical articles use more AI than this policy suggests, but I’ll disclose that in the article that does so.

There you have it. No mystery here, you don’t need to wonder if I’m using AI. I am, but in a limited way that leaves my work on this site as my own.

Now, just for fun, I asked a few different AI models to “Please improve the readability and formatting of this article”. Looking at what it returned (see below). I’ll admit it: AI is far more readable and accessible. It’s not even close. Content may be king (and I think I’m good at generating educational content), but without an easy-to-digest presentation, nobody will be interested in seeing the king. It also hurt a bit to read the “Major Flaws Corrected” section. Here is the output from multiple AI systems (note that I would normally take the output and proof it, rewrite parts, add things, and make sure it tracked what my original writing conveyed very closely, as well as some specific things I’d correct in the o-1 Pro, including changing “All of my videos feature me on camera” to “All of my videos feature me and my original content on camera:” and changing “However, I sometimes instruct the AI to include specific points related to provisional patent filings, such as” to “However, I sometimes instruct the AI to add specific points that were not in my original video transcript. For example, in talking about provisional patent applications, I would instruct the AI to add text conveying the following:”). You get the idea. Even the top of the line (or at least cost) o-1 Pro model needs a somewhat heavy hand in editing to keep it true to my original writing and content.

Added June 30, 2025: OMG I hate that the em dash is now taken as rock solid proof of AI writing. When I do drafts inside of Word, it automatically converts “–” to “–”. I’ve used the em dash for years, as shown in this 2011 blog post. Lately I’ve found myself changing the grammatically correct “–” to “-” or “–” simply to avoid people thinking that AI wrote it. I’ve had time to reconsider what I’m doing, and I’m not longer going to dumb down my writing to make it look more “human”. Besides, if I run something I wrote through an AI asking it to “please spot any grammar errors or logic errors”, I want to be able to make the corrections without worrying that people will think I said “please take what I wrote and create an article”. So — I’ll be using the em dash. Or, put correctly, “So – I’ll be using the em dash” (although past articles might not include it, and if I’m drafting in WordPress, there is no automatic conversion and who knows, I’ll likely miss many uses of “–“, which is how my writing has always been).

Added August 19, 2025: Wow. This sucks. Google’s search engine has indexed only the home page and no other pages on this site. I suspect (and Gemini agrees) that this is partly because using AI to clean up the writing is causing Google to pick up a signal that the content is low quality. In other words, the site is being punished by having AI contribute primarily stylistic changes because it makes it appear that the content is AI generated. As a result, I am rewriting many of the posts to revert to my “normal” overly dense and complex writing style. I’m sorry. I wish I didn’t have to do that. One thing I love about AI is that I can write and focus almost entirely on the ideas and story, knowing that I can later ask ChatGPT to “modify the writing style to be easily understood by people with an undergraduate degree” or “I wrote this for adults, but I realize it is great for kids too, please modify the writing style and word choices to be more comprehensible to children ages 10 to 14”. Those options are off the table until Google figures out how to properly deal with AI edited content, differentiating it from AI generated content.

Added January 13, 2026: It just keeps getting worse. It appears that Google still thinks my entire site is AI and isn’t indexing it. I decided to figure out what is going on. I purchased Grammarly and had it review articles for AI content. This is interesting: Things that I wrote entirely on my own were flagged as likely AI. It turns out that very precise writing is a flag, as is using certain phrases that are part of my usual writing style. Yay! It turns out that I write like an AI. Great. I’m glad I’m no longer in school, since I’d be failed for using AI when I’m not. I’ve had to “dumb down” my writing style, lose the precision, and change common phrases I’ve used for decades in favor of worse phrases — all in an effort to convince Google that what I wrote by hand was actually written by hand. I’m also adding in useless and distracting personal notes because those help score it as not being AI-generated. I remember when I was at UCLA and finally figured out how to avoid writer’s block — by simply getting what I want to write straight before starting to write. See? Irrelevant, distracting, but clearly not AI. I am dumbing down my writing so I don’t get penalized for writing in a way that Google thinks might be AI-generated.

Added January 21, 2026: This site is entirely unindexed right now (except for the home page) on Google. I’m almost sure that it is because I used AI to help rephrase things for clarity in the way described above. I’m now rewriting everything in a “dumbed down” way because precise writing triggers AI detectors, certain terms (“The Innovation” as in “The Innovation Cafe” is one) come up as AI generated, and it means that my content doesn’t get indexed or seen. What really sucks about this is that avoiding false AI positives means that I’m forced to reduce the quality of my writing. For example, like totally, I find myself adding colloquial phrases and I is sometimes adding grammar errors because that makes it more likely that I get indexed. Nice? Not really. Critical Side Effect: I am deleting the AI-written sections below — the ones I included as an example, fully disclosed, of how AI can convert human writing into something more accessible — because they will interfere with my page ranking.

The examples of AI writing that used to appear here have now been deleted — to avoid having Google think that AI is writing these articles and refuse to index the site as a result.