Struggling to get reimbursed for work travel because the digital font I signed my receipts from isn't sufficiently curly. I realize we are all-hands-on-deck to save higher ed...but at least some of this is self-inflicted.
Thomas F. Varley
@thosvarley.bsky.social
Dual PhD: Complex Systems & Computational Neuroscience - Postdoc at UVM in the Vermont Complex Systems Institute. Information theory, synergy, and emergence. Connoisseur of collapse phenomena
Me getting into somatic experiencing in 2023: "Wow, what an interesting idea that I can use to explore the embodied aspects of my chronic pain." Me now: "Wow, maybe this will help me hold on to some final shred of humanity in the face of the fizzing digital fungus filtering out of every screen."
I see this take more and more - was it a commonly-articulated definition of art before AI really got cooking? I have a hard time imaging that "quality is irrelevant" would be something most historical artists would have cosigned.
Gordon McAlpin@gmcalpin.bsky.social · 2w agoAI can't make "art," by definition, so i think you're wrong there. art is the expression of human skill and creativity. the quality is irrelevant; the intentionality and initial decision-making does, and computers have none. humans can ask an AI to generate images and pretend it's art.
Struggling to decide how I feel about this. On one hand, I will always believe that education (even in "useless" fields) is fundamentally important for being a well-rounded human. Otoh, I see what student loan debt has done to my cohort and something *has* to change. newrepublic.com/post/213446/...
Trump Moves to Cut Off Student Loans for Tons of Degrees
Under a new rule, if your degree doesn’t make enough money, you won’t be able to get a student loan.
newrepublic.com
This whole discussion seems very silly to me since humans are compulsive anthropomorphizers. We see faces in clouds and get grumpy at our computers when they take too long to buffer a video. There is just no universe where we don't anthropomorphize something that actually talks to us.
Ted Underwood@tedunderwood.com · 3w agoThis semantic question is going to get resolved in the way semantic questions deserve to be resolved — arbitrarily, by the actual practice of people who have to use English words to be understood. They already say "it's still thinking" and "it didn't understand me." They will keep saying that.
The collapse of MAGA/Trump's popularity won't lead to people sheepishly shuffling back to the center. Instead, we are going to see a Cambrian Explosion of psychotic ideologies that will finally be able to proliferate now that the choking shade of Trump's shadow is finally withering.
Will Stancil@whstancil.bsky.social · 2w agoIt's absolutely insane to be doing this kind of "Why I've gone MAGA" coverage at a time where Trump is basically less popular than any president in history at this point in their term. Completely out of touch with the average American. Bubbled elite nonsense.
I really want to know what made 4o a genuinely SCP-level cognitohazard. I hope they've still got the model weights in a lead-lined box somewhere at OAI because cognitive scientists need to study this thing.
Louis@louisevans.bsky.social · 3w agoNPR says it was 4o
Introducing Ferref - a reference manager for the AI age. github.com/thosvarley/f... I've been thinking a lot about how scientists should interact with AI, and one thing that really bothered me was that a lot of library tools are clunky, GUI-based interfaces that aren't great for agents. 1/N
Opus 5.5 can solve basically any information-theory question I want it to. I can get the answer to every puzzle that's ever bugged me throughout my career, with zero effort or friction. Turns out...that kind of sucks. Faust was totally right.
Seeing this just days after learning that 64k people submitted to ICLR makes me despair for the future of academic science. Other early-career researchers - how are you thinking about all of this? What are we doing here?
Iva Vukušić@vukusiciva.bsky.social · 3w agoThis ridiculous number of applications is, in my view, the result of 2 nightmare developments happening concurrently 1.the precariousness of the job market for young PhD graduates which then fuels further: 2.the AI tools enabling overproduction of proposals. #Research #ERC #ResearchFunding #Academia
I get myopically bogged down in annoyance with Claude for doing one silly thing or another, but then I step back and look at plots like this and think...yeah we may be into the foothills of Singularity.
Epoch AI@epochai.bsky.social · 3w agoAI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cost has fallen ~47%/quarter since 2023. That’s 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity.
I live ~30 mins south of Greenfield and have spent almost a third of my life in the Pioneer Valley. This is the most West MA thing I've ever read, right down to the clown school thing. www.bostonglobe.com/2026/09/22/m...
A Western Mass. lesbian bar required KN-95 masks. When it changed the rules, all hell broke loose. - The Boston Globe
Last Ditch Bar in Greenfield wanted to build community. They got a firestorm instead.
bostonglobe.com
Okay so @defector.com has had an odd glitch since the site redesign a couple of weeks back where the wrong photo and headline are often put together on the main page and I was going to ask them about fixing it but this just happened and as far as I'm concerned they have to leave it as is.
Do we even know if OAI told their model try Cordoba-style forcing, or did they just say "Solve Navier-Stokes, make no mistakes" and the swarm independently self-organized around that plan of attack?
The world actually IS ending...in a way. We're not all going to die, but "the world" as a collection of narratives, institutions, and norms that we use to scaffold our lives and make sense of what happens to us is absolutely falling apart. This is only partly due to AI. Who knows what comes next?
Nathan Lambert@natolambert.bsky.social · last mo.In light of kind of insane AI safety discussions recently: 1. AI progress is very fast 2. we should be careful about how we roll out the tech 3. the world is not actively ending
It's like Anthropic and OAI are in some kind of bizarre competition; can Anthropic provide service so abysmally bad that I'll eat my principles and go to OAI...or will OAI act so heinously that I'll stick with Anthropic in spite of their endless parade of BS?
It occurs to me that all the drama around the Navier-Stokes result is a tremendous stroke of luck for militant AI-antis: rather than acknowledging the reality of what OAI's models did, they can just say "the model just stole human work" (regardless of if that's even true or not).
It did though. The discovery that computation could be embedded in mechanical systems was a foundational insight that spawned whole new branches of math, new philosophies of mathematics, and the whole universe of computers. The lack of scientific literacy (or curiosity!) is kind of shocking imo.
This feels tragically reflective of a lot of thinking among vocally anti-AI people. Math is as much a human artform as poetry is - and is just as worthy of an existential crisis. But these people don't respect or value mathematics as a human endeavor.
Hellscape Navigator@hellscapenavigator.bsky.social · last mo."Computer Good...at Math" is not the basis for an existential crisis IMO if poetry turns out to be Math ill re-evaluate panicking
What drives me crazy is that these are all good and important goals...but these "stochastic-parrot takes" are actively harming that goal by refusing to acknowledge the obvious reality of what AI can do. How can you have good regulation if you deny the existence of what you're regulation?
Emily M. Bender@emilymbender.bsky.social · last mo.... rather than protecting our rights: to clean air and water, to privacy, to good education, to healthcare, etc. And rather than doing something about the massive wealth inequality all of this is also driving. >>
This is a great example of a particular kind of academic brain driving me nuts - rather than talking about what actually happened, instead the admonition is to instead *talk about how we're talking about it.* Meta-discourse gives enough recursive distance to avoid facing an uncomfortable reality.
Emily M. Bender@emilymbender.bsky.social · last mo.And as for fantasies of "AI civilizations" independently "hacking" into various companies' networks: it's worth asking -- who is describing it that way an why? In what other field would the companies perpetrating cybercrimes brag about it? >>
I predict that as AI accelerations leads to increasingly widespread disruption, the boundaries between the reactionary left and reactionary right will dissolve and become a huge blob of psychotic anti-modernity nostalgia. Whether it's batteries, vaccines, or trans surgeries, they will hate it all.
David Roberts@volts.wtf · last mo.I know we're all supposed to be herding onto the anti-data center bandwagon, but man, I worry a *lot* that it's going to overspill its borders & become a more generalized anti-wind & solar, anti-battery, anti-grid-infrastructure, anti-transmission, anti-anything mega-glonzo that leaves us worse off.
New #preprint! Led by Elena Sergeeva. @pigozzife.bsky.social @thosvarley.bsky.social Colin J. Comerci Zhe Zhao Robert Brucker Gürol M. Süel @doctorjosh.bsky.social www.biorxiv.org/content/10.6... "Living multicellular systems induce decodable spatial patterns in bacterial collectives"🧪
Living multicellular systems induce decodable spatial patterns in bacterial collectives
Living systems continuously modify their environments through chemical, mechanical, metabolic and bioelectrical activity. Whether a presence of a multicellular system can be encoded into the emergent ...
biorxiv.org
I've grown to *hate* the term "accountability". You should be accountable to your friends, your family, and your community. There's absolutely no reason I (or anyone else) owes "accountability" to anonymous randos merely by virtue of shared identity characteristics or whatever.
web_rant@webrant.bsky.social · last mo.Bluesky is a community of users. When a community rejects a user account that tries to promote hate or dangerous behavior that community responds accordingly to protect its people. Do not confuse accountability with "harassment". bsky.app/profile/safe...
What I want is a feature where, if your comment is too similar to one already in the thread, it doesn't get posted and instead you are invited to upvote the post that beat you to the punch. It would massively reduce dogpiles and also the overall level of redundancy/hive-mindness on here.
Bluesky Safety@safety.bsky.app · last mo.Dogpiling or harassing users is unacceptable. We have taken recent action against accounts for this behavior, and we will be increasing our efforts, consistent with our Community Guidelines.
Any tech billionaires want to help me start the Pioneer Valley Complex Systems Research Institute? I'm semi-serious about this: as Academia grinds against the rocks of demographic collapse, loss of public trust, and costs, we need new places for people to work on complex, high-dimensional issues.
This is really sad - it sucks to see someone's awesome passion project get torn down by maniacs. Can you imagine the world that these people would create if they actually had any power? It'd be a nightmare.
Rick Brewster@rickbrew.bsky.social · last mo.Alright everyone, message received. I was not expecting this response, I honestly had no idea before I made the post Like I said, the code is experimental. A proof of concept showing that PDN-on-WINE is possible. It's not finished. I can literally just delete it. It won't be part of the next build.
This is super cool - for scientists interested in reproducibility, I'd highly recommend learning Nix.
Isabella Velásquez@ivelasq3.bsky.social · last mo.Tired of "it worked on my computer" breaking your #RStats projects? Tomorrow on the Data Science Lab, @rpodcast.bsky.social is going live to demo {rix}, an @ropensci.org package that uses Nix to build reproducible R environments. I caught a sneak peek and it's WOW! Join us! https:// pos.it/dslab
