Ivan Flis

Imagine that.

Neovisno o tome sto mislite o Thompsonu i ustasovanju, svaki gradanin Zagreba koji glasa za HDZ nakon danas pljuje u lice svojem gradu i sugradanima. Mi ne zivimo u drzavi koja ima ustavni poredak, na milosti i nemilosti smo Andreja Plenkovica.

Zato je Zagreb besraman: bsky.app/profile/mmjs...

Apologies for all the typos, no edit function and foggy morning brain conspire. At least you know I wrote it, eh?

If the relationship to language is social, and if there is a machine that abstracts our language only to re-inscribe it, then that relationship-to-language becomes a “thing” — a manufactured commodity —that overshadows our relationship to language as a participatory social activity

Advancing regulatory variant effect prediction with AlphaGenome - Nature AlphaGenome, a deep learning model that inputs 1-Mb DNA sequence to predict functional genomic tracks at single-base resolution across diverse modalities, outperforms existing models in variant effect...

I had intended to post something about this new Google DeepMind paper that appeared yesterday in Nature, but the press coverage has added to what there is to say. So this is a long 🧵
www.nature.com/articles/s41...

Artists trying and failing in representing the past tells us something new both about the past and about the present. Why the fuck would we want to automate that?

What the slop does though is cut out the inventive artistic vision of the past that reinvents it for the present. It gives us a collage of already regurgitated representations drawn from the data it was trained on, remixed not as a pastiche, but a shallow rearrangement.

No media production is an accurate reproduction of the past, some are slightly better, some are slightly worse, but they are always motivated by what our values make us look for in the past.

Which leads me to the 2nd point. It’s not that AI slop about historical themes like Aronofsky’s misses where other productions don’t.

They criticize the public images in order to remind the publics that they do not sit out of time, our visions of the pasta are current imaginations. You see whiteness because you expect it, not because it was really there.

This always puts the experts on history into a revisionist position - both toward previous academic reconstructions of the past and the the past framed by contemporary mass publics.

1st: the different political inflections of the past in popular imagination are ideological products of schooling and previous media exposure, but a crucial aspect of what makes them successful is that they are invisible to those who believe them. (“all Romans were white, that is historical fact!”).

Now, what is the undercurrent here is that all those who believe the past was not multiethnic are racist (I even implied it up there) and that media can offer us a true picture of the past.

Mary Beard is right, Roman Britain was multi-ethnic – so why does this upset people so much? A row about whether Roman Britain was ethnically diverse has turned nasty.

I am thinking about the episodes like British historians clashing publicly with the white nationalists, racists etc. who had a problem with the idea that media try to represent some aspect of how, in our terms, Roman Britain was multiethnic theconversation.com/mary-beard-i...

Before people started using the recently develop tools we call AI to churn automated art replicas, our publics have been clashing about the “true” representations of the past.

Aronofsky made an AI slop historical series about American history. People are going online to make fun of it and criticize it for its ahistoricism and it got me thinking abou our sense of ahistoricism and mediatization, and how twisted up it is and the AI slop exacarbates it.

Interesting thread that touches on issues I have tried to address in my articles about replication and well as my book (especially chapters 7 and 8).
#psychology #philsky

Ahahahah. Mladen na ozempicu je frontmen Ragea!

My impression is that some people are taken aback by some thought experiments becoming real world problems with moral and political consequences. Some are taken aback, others are riding that gravy train like there’s no tomorrow.

via bsky.app/profile/aelk...

Reminds me of 19th century spiritism and esotericism. It was also buoyed by fantastic technologically induced speculation about electricity, aether, energy, inhuman intelligence and compelling fiction.

The communities produce their literatures for a purpose, however opaque it might be. I’m fully willing to admit I am missing the mark, but I think so are the takes that see this state of the literature as anything more than a symptom.

Those things triangulate the lodestars for the communities, and they overproduce similar measures of the supposedly same thing. It’s not a problem that needs to be fixed (we need less such production), rather, a set of communities that need (or need not?) orientation..

This is the thing, that there is a convoluted process of many interactions in how publishing works, what people are trained in as good methods of inference AND description, and what the nominal goal of scientific production is.

So, what’s a better candidate for a community-wide motivation in generating all this variability? It can’t be that people are dumb or untrained IMO.

Yeah. Give me a detailed, well-thought out description over a snazzy conclusion anyday. The inferences, even when warranted and good, will have a much shorter half-life than good, work-intensive description.

But that’s the thing - what (I think) motivates the community is not solidifying the measure for all, but extending it all the time through (more or less) careful operational analysis. What would it spell for their research communities if they used it consistently?

(I’ve been reading @ufeest.bsky.social new book on operationism with my students this semester, this is where a lot of this thinking is coming from)

Yeah. That’s the thing, is that most of psychology is actually painstaking description dressed in the trenchcoat of NHST and hypothetico-deductivism. But how those two motivations interact is where the problems arise IMO.