2019
The problem with AI is that most competitive grounds are already simple enough to be solved more efficiently with learning algorithms.
SourceAI systems
Recently, I built conceptual software about AI.
Evals, containers, harnesses in different shapes.
I concluded that:
Simplicity takes
so much effort.
AI notes, 2019–2026
I took these
possibilities seriously.
2020
AI will generate most popular art and entertainment by 2030.
Source2021
We'll be reviewing and debugging AI-generated code rather than writing code on our own.
Source2023
Crypto and AI are two sides of the same coin.
Source2026
I don't fully understand guardrails. A large language model is like… a semiotic star, in all its explosive projections. A guardrail is like a tent above your head.
SourceWhat follows?
What changes when
sounding right becomes cheap?
I kept writing about
the impact of AI.
My bet is an
epistemic recentering.
Changing the frame
nothingness
The only thing
absolute is
absence itself.
Non-absence (reality) simply expands
and differentiates
indefinitely.
I explore this in mathematics, too.
Music and writing
I make music.
I've kept at it, sporadically,
since I started playing guitar.
I write.
A company, described plainly, is a claim: that value sits in some region of possibility, and that this company can reach it better than anyone else. That region is its real product. Call it the search space — the ground. Ground of this kind is finite and rivalrous; two firms cannot hold exclusive title to the same plot. So a company is, in the end, the working of a search space for as long as it can keep others off it.
From The Ladder Is a ClockMy family
Happily married to
the love of my life.
Father of a boy
named Oliver.
The name
At 10, I wrote a book
that I deleted.
The main character was named after
सदसत्
sadasant.
References