4.11.25
legibility vs. creativity, collaboration is not a product, personalization in AI, micro as antidote to macro
So this is week one of actually writing these in a substack and I find myself oddly needing to write some pre-amble up front. Saying I feel the need to write one counts in fact as a pre-amble doesn’t it? Anyways, it’s been a weird week. I feel both deeply excited about the wonders of the world and deeply uneasy about the uncertainty around us. So with that, my weekly musings…
Legibility as antithetical to creativity: Legibility in many ways is antithetical to creativity. Our drive to streamline, to TLDR, to summarize, reduces the process of how things evolve and come to be. The process and the re-working of something, the art of tinkering can drive beauty and maximize creative output. Social media and the drive for immediate response (do what maximizes likes vs. makes you feel good) was already a huge destructor. AI in this way poses an existential risk to creativity and true discovery (which I would argue is creative). I've always been a weird internet kid. I loved the creative alchemy of disaggregated internet searching- going down a rabbit hole unsure of where it will take you, often resulting in a larger unintended web of thoughts. AI is meant to give you the exact, precise answer you are looking for. The ability to quickly get to an answer, to a singular response reduces the need for process and thereby the opportunity for serendipity in a thought or a creative process. This is a bit of a circle back from 3.28.25 but I’ve just been thinking a lot about creativity in the world of AI. (s/o to Jackson Dahl’s interview with Nabeel Qureshi for getting me thinking on this one. You should read/listen- so much packed in)
Collaboration is a feature not a product: I don't believe in "collaboration"as a core product. It can be a valuable feature of a broader platform but any time I hear "easier to collaborate" as the main value, I am immediately doubtful of the ROI that platform will drive. People don't pay large sums of money to "just" collaborate IMO. You have to drive cost savings, time savings, or drive revenue.
Adaptability/Personalization as a core limitation of AI: current best in class tools like Claude and Cursor struggle with adaptation/evolution given the limited context windows. You can hack this with a memory layer (like OpenAI recently announced), but the model itself isn't actually adapting to you as the individual user. You can prompt a tool to create a new output and have that stored somewhere and then selectively retrieved (from my non-technical understanding), but it's not truly adapting to you as the individual user. There are smart people (namely one I spoke to that generated this thought) who are working on enabling models to adapt to new problems and domains they haven't seen before (vs. using a memory layer). If you could do this, you could imagine a world, where every individual had their own custom model running on their own devices - structuring the world in a way that was unique to them. I'm super fascinated with this concept, so if you are working on anything around this, I'd love to riff more. (s/o to PhD student- leaving you anonymous so folks don’t bother you, unless you say otherwise)
Day-to-day as an antidote to macro-whiplash: Start-ups and namely the extraordinary intensity and focus required to operate one day-to-day is a haven (or perhaps hell!) from the macro chaos. Spend too much time thinking about what will happen at large and you can go crazy. Staying focused on creating value relative to a problem feels like a really nice respite at the moment. Nice to focus on the micro vs. spending time too high right now.
Lastly a quote I heard this week that I just loved as it related to building companies: “I view the world through the lens of momentum and leverage." (s/o to anonymous unless you too want to be quoted!)
And a song I discovered that I thought was fun (and really loved the title of): the folly of wisdom by the olllam
Stay weird. Stay curious.
-CB