UPenn CS + Robotics
The only way to create something profound is by creating something related to the thing(s) one obsesses over. Further, one can only really obsess in select specialized skills. To find things to specialize in, you must pursue the things that are authentically you.
In other words, we find: Authentic, obsessive pursuits → profound creations
This then emphasizes the importance of creation as a daily habit until creating is as natural as breathing. Whether the medium is coding, writing, etc. these pursuits will allow for fundamental creative development. Hard skills do not come easy; persistent effort in creative outlets will eventually yield exceptional results.
Focus on a few things and create things while learning. Keep doing this and something meaningful will happen.
We conjecture that people primarily scale skill sets for which they have a natural affinity for. This suggests an adaptability to a wide range of domains could be a rare skill set; especially orthogonal skills in what could be thought of as the "space" of those an individual can possess. Note that I am not referring to any particular domain but quite literally any skill set.
Continuing, we consider what skills have the greatest return on investment to learn and more specifically which are congruent with a person's goals. Some combination of skills (based on some multivariate metric composed of "orthogonality", "congruency", etc ) could lead to a truly holistic, rare, and hard-earned skill set.
A lot of people only develop things that they had a natural knack for and are afraid to stray far from what they know. Therefore, a willingness to stray and build deep technical skills for things that someone is not readily adept at gives someone a unique skill set.
Human-human connection → Human-robot connection → Robot-robot connection.
There is this notion of the "dead internet" theory. Robots infiltrating our online channels, running rampant pushing government agendas. The idea would enrage most people: these are echo chambers but they're our echo chambers.
The transformer architecture and internet scale data has placed us a few steps away from this future. But in a turn of events, it does not (readily) appear that there are overarching figures creating large scale tomfoolery, but rather individuals creating local misuse.
Reduce quantity, improve quality, and attain a willingness for distinction; as the gap closes and these models become increasingly cogent, we can keep our communication authentically human by allowing ourselves to stray a bit out of distribution.
LLMs are becoming increasingly good at emulating human text. We can combat this by allowing ourselves and the way we write to deviate from the norm and be far less replicable.
Orientations, the direction of pursuits, are likely one of the most important things to get right in life. This may seem trivial; obviously orientations matter. They define a career, relationships, etc. They are also quite individualized and context dependent.
It appear, however, as if everyone is orienting themselves in fairly close alignment with each other. I conjecture, this stems from an unwillingness to trek into an unknown. Sticking to the conventional path can provide a crippling positive feedback loop. A feedback loop which admittedly leads to a generally positive outcome in expectation.
Walking along the conventional path or pursuing an entirely different one is not a binary decision. Sometimes it just takes a willingness to orient oneself in a direction uniquely one's own and take one step and another.
It's easier to do your own thing than you think. It just takes some small step.