My data and a new modality of thought
Regrets on not having a significant corpus to train on, how to think about the data relevant to this newly introduced modality of thought and forward-looking musing on a symbiotic relationship with AI
I’ve always considered myself a writer, perhaps not a great one, but a writer, nonetheless. I think there are several reasons for my love of writing:
I’m a voracious reader, always have been since the somewhat late acquisition of the ability around 8 years old. Encyclopedias, reams of novels, thoroughly bathing in the Great Books, eventually earning BAs in Classics and Philosophy, a minor in English, an MA in Philosophy and a JD. Much reading.
With writing one can cheat time and space. My friends span millennia, you’ve met them too regardless of whether you’ve ever travelled outside the lintel of your own home. I too can cast myself over miles and beyond death by tapping out these words.
Language itself is delightful. Have you noticed the feeling you get when you have the insight as to precisely the correct word to use in a given context? Have you experienced having the right word on the tip of your tongue but not having the insight, ruefully using a synonym that holds a poor candle to insight you wished to convey?
These three mighty reasons allude to a fourth, and for the purposes of this hastily written post, the lynchpin of the whole love of writing endeavor:
“Writing is thinking. To write well is to think clearly. That's why it's so hard." - David McCullough
Writing requires a secondary insight of formulation, that is, the articulation of your original insight in words that can convey that insight to another mind. To do this, we must be clear with ourselves what our insight is and what context is necessary to come to the same insight, which predicates an understanding of our audience.
Now, imagine a triangle composed of you, the individual to whom you hope to impart meaning, and the mutually understood thing. The mutually understood thing is in a continual state of change throughout the dialog until both you and the other individual have determined through mutual agreement that the formulation was successful in conveying the insight.
This is a speedy process if the content is simple (Apples are good to eat) or the context rich (we both have PhDs in early medieval European history and concur that legumes were indeed vital to the artistic and technological achievements of 1000 AD), but can be very long and complicated, as our robust education systems can attest.
Why does all this matter in the Era of AI?
Well, because I want to give it all the context I have. Here’s a “mind”, a little engine that could, and I benefit by giving it vast amounts of context if it’s going to do anything other than create pretty, simple, random images or tell me “Apples are good to eat.” I have two ways, currently of enriching the context of this little engine:
Giving it as much context as I can through adding custom instructions, prompt engineering like a boss, increasing its context window so I can pump whole pieces of work into its bounding box, and otherwise providing what I myself would need to make intelligent decisions or come to better insights around this problem set.
Asking it to pretend like it has the necessary context (through prompting) by telling it behave or respond like someone who does have that context.
Great, both are helpful, initially at simple things, and slowly at more complex things as we ourselves get better at our secondary insight of formulation for this particular “audience” and as the context window continues to grow with our technological developments.
But where I get incredibly excited is the pursuit of AI as a “modality of our own thought,” an engine that has all the context I myself do.
Imagine an engine that has the known universe in scope but can convey it to me in ways that I can understand best or help me come to better insights because I’m asking better questions, or giving me the perfect recall necessary in the moment.
I’m assuming this is possible with more context, so how do I think about my context?
Assembling and Creating Data for better UX of the Modality of Thought AI
Well, what do I got:
Biographical data: a biography of myself, including third party data sources: geographical location, school events, extra-curricular activities… etc.
Work Product: school essays, published work, work deliverables,
Genealogical data: thanks 23&me.
Bio data: Garmin, Oura, periodic bloodwork, diet, exercise.
Relationship data: everyone I’m friends with on any social platform.
Phone tracked data: location, a few other things.
What do I wish I had?
A list of every book, article, or piece of published writing I’ve ever read.
A list of every movie or episode I’ve ever watched.
A CRM with relationship metadata for every relationship I’ve ever had with any human or animal ever.
A transcript and video of every conversation I’ve ever had.
A video of everything I’ve ever seen or heard.
A transcript of my inner dialogue over the course of my entire life.
What can I reasonably get?
I’ve logged my browsing history for years and could likely come up with an 80% comprehensive list of books and articles from my coursework and Feedly archives.
I could likely get 80% here too, just listing out things I recall and bouncing them off a few dozen close friends.
Is this one awkward? I do have a personal CRM, should I go through and metadata up every relationship?
Super impossible, but I can add in OtterPilot AI to every virtual meeting I can. What do I do about in person conversations?
Also super impossible, but maybe I get those Meta Ray Bans…
This last one is the reason I regret not writing more.
A transcript of my inner dialogue over the course of my entire life.
While this is impossible, it highlights something that could have been possible (and still can be going forward). I could have driven more of that inner dialogue to clarity through a dedication to and practice of capturing secondary insights of formulation.
In my innovation practice I preach the discipline of documentation and the rigor of review, both of which I should practice more in my personal life. The Modality of Thought AI will open a categorical shift in the way we as humans can pay attention, reason, and decide. But only if the data is there.