Maxim Raginsky
@mraginsky.bsky.social
web: http://maxim.ece.illinois.edu substack: https://realizable.substack.com
Wise words from Richard Sennett's _The Craftsman_: "The enlightened way to use a machine is to judge its powers, fashion its uses, in light of our own limits rather than the machine’s potential. We should not compete against the machine. ...
Came across the well-known quote of Oppenheimer in a book: “When you see something that is technically sweet, you go ahead and do it and you argue about what to do about it only after you have had your technical success. That is the way it was with the atomic bomb.” That's the way it is with AI now.
where have you gone, Herbert Simon, a nation turns its lonely eyes to you etc etc.
Y’all better be aiming at the good life, with and for others, in just institutions. Just sayin’.
I need to start carrying Cioran and Eliade in my bag alongside this book, for the full Romanian immersion.
Quine's indeterminacy of translation obviously applies to the process of translating mathematical texts to their Lean formalizations. Useful to keep in mind in the context of all the current AI-and-math discourse.
In this episode of @braininspired.bsky.social, @mraginsky.bsky.social draws on control theory, engineering and philosophy to explore what makes biological brains intelligent, what AI is missing, and what principles underlie biological autonomy. #neuroskyence www.thetransmitter.org/brain-inspir...
How does a control theorist explain cognition?
Maxim Raginsky draws on control theory, engineering and philosophy to explore biological and artificial intelligence and biological autonomy.
thetransmitter.org
This is a bit all over the place, but: fried longanisa sausage, okra in tomato sauce, dirty rice.
“[H]e did not mean to imitate those philanthropic models who make a profit out of poisonous pickles to support themselves while they are exposing adulteration, or hold shares in a gambling-hell that they may have leisure to represent the cause of public morality.” (about Dr Lydgate in _Middlemarch_)
Ricoeur on reading, very relevant for LLMs that act via the interface of language: "reading ... is at its best a struggle between two strategies, the _strategy of seduction_ pursued by the author in the guise of a more or less trustworthy narrator ... , and the _strategy of suspicion_ pursued ...
“All prediction is conservative: it demands that we be as we are in whatever future it constructs.” — Paul Valéry, “Unpredictability” (1944)
trip down memory lane: Pie ’n Burger, Pasadena, CA, 10 years ago
“It will perhaps one day be necessary to forbid actually doing what today science fiction is limited to dreaming about. But have not dreams always transgressed into the realm of the forbidden? So let us dream along with Parfit. …
Making casarecce con le sarde tonight, paired (appropriately) with a Sardinian white wine.
I find the "digestion" metaphor (i.e., mathematicians "digesting" AI-generated proofs) vulgar and off-putting. It's a sign of the times, genuine intellectuals like Yuri Manin are being replaced by a new generation whose identity is shaped by narrow professionalization and impoverished imagination.
this week’s Negroni proposition 1oz Roku gin 1oz Oka Kura sweet Japanese Bermutto sake vermouth 1oz Select Aperitivo
here is a succinct thing I can say: "alignment"-style AI safety risks DO NOT EXIST or need remediation with LLMs as such (there are lots of other problems) and OpenAI and Anthropic are in the business of TRYING TO MAKE THEM EXIST so that they can remediate them.
tweety fish oh wow like a chimera or something spoooky@sifu.tweety.fish · last mo.it's so funny how they imported all of their concerns about alignment and internal motivation to an algorithmic framework that... really doesn't have those things, architecturally.
“established … by a number of savants (mostly from the Soviet Union)” — LaSalle and Lefschetz, “Stability by Liapunov’s Direct Method,” 1961
I see Bluesky is now speedrunning the intentional stance and multiple realizability. Okay then, carry on.
No one should be taken seriously when they say things like “update my priors” unless they can explain how to derive the Kalman filter. Desperate times call for desperate measures, people!