Federico Adolfi

After dislodging the meteorite fragments from the roofs, people sat the fragments at the kitchen table and gave them a speech about being late to deliver signs of the apocalypse

An empty stage prepared for a jazz concert

I went to a really cool jazz concert a couple of days ago but no photos or recording of any kind were allowed during the performance so here you go, social media overlords #jazzsky

Screenshot from a chapter by Wimsatt reads: “A similar point is useful in the analysis of complexity: systems for which these different perspectives yield decompositions of the system into parts whose boundaries are not spatially coincident are properly regarded as more descriptively complex than systems whose decompositions under a set of perspectives are spatially coincident.”

Meanwhile

That’s very cool. Well give me a call if you ever need a…scientist who dreams of your exact academic unit 😂

Everyday I dream about a genuinely interdisciplinary cogsci academic unit where you can spontaneously ambush anthropologists, archeologists, etc (besides comp sci, neuro, etc) to do quirky multidisciplinary projects

Upcoming talk in the Theory of Interpretable AI seminar: Federico Adolfi frames interpretability in terms of computational complexity theory.

Kid hits their sibling? Hit your spouse to show them how wait no don’t this I did not think this through

Me: I’m elated to announce my new suite of agentic LLMs fine-tuned to exhibit subtle behaviors that maximally elicit complex vicarious emotions in humans such as shame and embarrassment

Billionaire: I can help you grow with an initial investment of—

Me: I call it CrinGPT

B: get out

I’m gonna build a model that is so theory neutral that it cannot do anything. It’s so over for you scientists

German: Information
English: gotcha, that’s information, next!
German: der die das den dem der des
English: get out

Yeah I get that. I guess specifically I’m confused about the unique contribution of reposting in the context of the multiplicity of feeds people are already exposed to. I get that I wouldn’t see a post if I’m not following the account and they are not on any of my feeds, but otherwise 🤷🏻‍♂️

The former: resposts in general 😅

Broadcasting my ignorance but I’m still not clear on what reposts are supposed to accomplish/contribute on this, the only site I’m on

Academics blueprinting: “3 weeks ago my team and I embarked on what turned out to be the most important journey of my career. Have you ever wondered how the brain does behavior? In this thread I will reveal the answer. What you’re about to see may disturb you. But first, a word from our sponsor…”

To take this to the limit, we note that for every ε-ssay there exists a δ-ebate.

In this ε-ssay, I develop a kernel of an idea so diminishingly small that you’ve just read it

You’re welcome

I wanna say… BBS article by Pylyshyn (80s). How did I do?

Selfishly paging @mraginsky.bsky.social because I too would like to know the answer 🤓

Me: alright then, time to work on the abstract for my talk and ONLY the abstract for my talk until it’s done

Also me: *spends an hour working on a schematic for a single slide* Nailed it

Thanks, currently recharging motivation and calories 😅

Lemon gooey butter cookies with powdered sugar on top

In this household we are generally self deprecating and also we celebrate rejections with dangerously delicious cookies

I don’t know who needs to hear this but

*uncanny mouth noises*

Oh hell no
Put. That. Shit. Back.

Oh hell no
Put. That. Shit. Back.

Got a little taste of that Academic Rejection yesterday and it already feels like a 27 year old unresolved trauma

I don’t know who needs to hear this but

*uncanny mouth noises*

Gotcha, yeah especially in ML where the significance of the details is largely unknown, and yet there’s a layer of wishful storytelling around them that goes forever untested

The work above might seem quite technical but my super secret personal motivation for doing it has always been these deeper (philosophical, some might say) questions and how they impact other downstream concerns around scientific modeling and alignment

One aspect I don’t see on your list or in the replies (& I’d be curious why, but I digress) is the issue of what it would mean for us to understand them; eg ability to *efficiently* answer a family of questions useful for description, explanation, prediction, control, as in👇
bsky.app/profile/fede...

Ha no I mean if you are saying you’re not interested in artificial systems, just natural organisms