Taking Tacit Knowledge and Making It Tangible

My current thoughts about AI and writing, as a writer working in marketing.

When any LLM can put out vast quantities of content that is 80-95% almost ready to go out the door/written in voice/tone for a company, why even have writers on your marketing team?

Creating "new data" is the most valuable thing a writer can do.

Conducting interviews with subject matter experts and take what's locked up in their brain and put pen to paper. Assumption: LLMs/AI have already been trained on all "publicly available data." While it's generally true, it's not 100% true.

This is why the idea of oral history/capturing family history and genealogy research is so interesting to me. That knowledge, if not recorded, is gone forever when your loved ones pass away. I like the idea of an app that can help preserve this information, but the actual idea is much bigger than just this one thing. It's about everything, not just family history and family stories. It's useful to organizations, especially for marketing strategy too. What types of content will you create when you can create absolutely anything you could think of? And why would you pull from the same library as all your competitors when the most interesting thing about YOU or YOUR company is not what is already publicly available?

It's because people are lazy.

I am lazy, too. The path of least resistance to get a goal accomplished makes the most sense and preserves the most cognitive resources and energy.

It is very slow to sit down and have a conversation with someone. Not everyone is good at sitting down and putting their thoughts on paper. I struggle with it too and it's all I do, every day. It's easier to open up the audio transcription and just talk. The simple act of making a record is all you need to do. But that is a big lift for some.

With the tools available today, whichever or however many LLMs/AI tools you use, in a few minutes, it can take your raw input and it spit out something 80-95% of the way there for you to edit and refine. Basically all steps in the writing process have been eliminated with the exception of raw brain dump data/extraction of information from biological data silos and quality control of the final product.

It really does come down to asking the right questions. Asking yourself, other people, and also AI the right questions.

If I were to write a personal memoir, (I wouldn't though) I'd compile the collection of everything I have ever written or created. I wouldn't worry about sorting or organizing it, just collecting it in one place. Browse the collection, add in net new raw data for things I wish I had of written about. Maybe make a short list of important or memorable events that absolutely need their own chapters or sections on. I'd think carefully about what is most important to share that could be maximally useful to other people. I'd think carefully about what I don't want people to know like my kids for instance, remove those bits of at least mark a chapter called "Unhinged, NSFW, If you're my child please skip this chapter (You'll thank me later.)" I don't like the idea of removing uncomfortable things because it's part of my story. A disclaimer will work just fine. I'd write it like I were already dead, very much "Who Cares?" (But I do, and that's why the naughty bits stay in and here's why: People are multi-faceted and terrible and wonderful and how can you actually appreciate the amazing parts when you don't know all the fucked up parts too? I have a bit of a boner for the truth too. That's a turn off for some people. Don't be afraid to ramble when you write. What you spit out is almost as interesting as everything else, besides, people get to glimpse your never-ceasing, annoying-as-fuck internal monologue.)

Then once I had that collection, I'd get AI to put it in a logical order, probably pull out all date metadata because putting things on a timeline makes little more sense. It's easier to follow and understand key decisions when you see where things started and where they ended up.

I'd use a tool like xmind to start a concept map and include that in the collection. This is also a manual step of taking out tacit knowledge and putting it in a tangible form. It's important because it also helps you highlight the questions YOU have about things that happened and it's interesting to see how you sort it. Probably helps you identify places you didn't write enough about also.

Anyway, I know I have some old writings on codifying knowledge. I tried to find them to no avail, and I have one minute to finish this before my job starts. Sometimes you just do the best with what you have. At least you tried and continue to try. Writing doesn't have to be perfect, especially if you want to present the actual truth.

This is very much not done and I have a lot more to say on the topic, but in the interest of time, this will have to be my record and my attempt at codifying my thoughts on AI, writing, and what humans should focus on if they want to thrive in this new AI for everything world.