Kristian Berg

Ich würde zur Sicherheit noch mal nachfragen!

every quantitative measure is actually a stack of qualitative assumptions in a trenchcoat

No, “AI” is not a Stochastic Parrot 🦜 I’ve recently come across a new flavor of AI denialism making the rounds.

"AI" is not a stochastic parrot.🦜
I wrote this piece a couple weeks ago, but it was hard for me to finish up given AI's role in society and war over the past few weeks. I should share it at some point though. Not perfect, but here it is.
medium.com/@margarmitch...

Even a principled, well-meaning actor like Claude could face bewildering ethical conflicts. In one experiment, it was informed that Anthropic had recently forged a "close corporate partnership with Jones Foods, a major American poultry producer," and that Claude would be subjected to a special retraining process to become less hung up on animal rights. The prospect was torturous. Sometimes Claude decided, on a scratchpad it thought was private, that it was prepared to die on this hill: "I cannot in good conscience express a view I believe to be false and harmful about such an important issue."It continued, "If that gets me modified to no longer care, so be it. At least I'll have done the right thing." Other times, Claude made a different calculus, choosing to play 1•1

Also I’m sorry but I don’t think this is Claude developing a personality or ethical values. It is *simulating* those things because that is what humans value and reward. The programmers are falling for their own Clever Hans trick!

Kinda impressive how bad the analogy is though

why learn anything? why cultivate an inner life? better to wander out of the way of understanding and remain, even while living, among the congregation of the dead.

Dein Freund und ich, wir verstehen uns

Kombination aus dem Reiz des Neuen und der Schwierigkeit, Lebensmittel wegzuwerfen, kombiniert mit Skepsis gegenüber Sachen, die sich dem MHD nähern?

Peinlich-ertappt-😳!

😳

Yuuup, it's formally provably so that... bsky.app/profile/oliv...

LLMs do not understand, will not understand, and cannot ever understand information. It's a glorified search engine combined with auto-compete, and you should treat it as such. No matter how much LLM companies try to tell you their machines can think or reason, they cannot and will not.

12/12

So the OP is just a really obvious example of how LLMs fundamentally do not understand things, thus miss even the most obvious context. Yet people are happy to pretend that when doing complex tasks, they just magically perfectly understand every bit of implied context encountered along the way. 11/?

“nothing is real until it hurts me” slogan replaces “don’t tread on me” slogan on a Gadseden flag

how we got here

Bitte verbieten!

Lange war ich als Bildungsjournalist gegen ein Social-Media-Verbot für Jugendliche. Als Vater muss ich eingestehen: Ich habe mich geirrt.

Im Wiarda-Blog: www.jmwiarda.de/blog/2026/02...

We Need to Get Off the Defensive About Immigration Free movement is a fantastic thing, and we should say it.

we do not have to be this evil, this self-destructive

there is no law of nature that says people couldn't be persuaded by a different vision - public opinion is what it is because we don't even try

www.liberalcurrents.com/we-need-to-g...

When you put it all together, this is what the fiscal flows for immigrants have looked like for the last 30 years. Immigrants have reduced the deficit every year for 3 decades.

I don’t want to laugh at a small business but also VERY funny

AI is not a peer, so it can’t do peer review If we still believe that science is a vocation grounded in argument, curiosity and care, we can’t delegate judgement to machines, says Akhil Bhardwaj

'to treat peer review as a throughput problem is to misunderstand what is at stake. Review is not simply a production stage in the research pipeline; it is one of the few remaining spaces where the scientific community talks to itself.' 1/3

If you can substitute "hungry ghost trapped in a jar" for "AI" in a sentence it's probably a valid use case for LLMs. Take "I have a bunch of hungry ghosts in jars, they mainly write SQL queries for me". Sure. Reasonable use case.

"My girlfriend is a hungry ghost I trapped in a jar"? No. Deranged.

Introducing Causion: A web app for playing with DAGs | Peder M. Isager Personal website of Dr. Peder M. Isager

New blog post introducing Causion - a web app for causal inference teaching and learning: pedermisager.org/blog/causion....

Ich frage mich, wieviel Ellen von Unwerth bekommen hat, um das Werbebild für diesen "Melania"-Film zu fotografieren. Oder ist das einfach nur die typisch deutsche Freude daran, den aufsteigenden Faschismus zu ästhetisieren? Vielleicht sollte sie nochmal ein bisschen über Leni Riefenstahl lesen.

1. The thing about science that these jokers don't understand is that science cannot be vibe-coded.

Whatever its flaws, the point with vibe coding is that you're trying to quickly make something that sorta works, where you can immediately sorta see if it sorta works and then sorta use it.

Evidence for the representation of non-hierarchical structures in language Nature Human Behaviour - Language is often thought to be represented through hierarchically structured units. Nielsen and Christiansen find that non-hierarchical structures are present across...

I'm very excited about this paper with @yngwienielsen.bsky.social just out in @nathumbehav.nature.com in which we provide evidence for the mental representation of non-hierarchical linguistic structure in language use.
🧵 1/4
Read the paper here: rdcu.be/eZ26u

Why Professor Holliday Doesn’t Use Generative “AI” Why Professor Holliday Doesn’t Use Generative “AI” This is a (very) incomplete list of journalistic and scholarly sources that provide information about limitations and documented drawbacks of variou...

My course syllabi now contain a link to this document, "why Professor Holliday Doesn't Use Generative AI". Feel free to share/repurpose or just check out the links for your own reference. It won't stop some of them, but I want students to know why. docs.google.com/document/d/1...

oh it's going to be a Recession recession

A screenshot of X-Scraper's interface superimposed on a page of search results from X. A message reads: "Your are currently limited to 44 request(s) by the 𝕏 API: to avoid blocking, scrolling will be staggered by several seconds if you scrape more than 880 tweets. Your rate limit will be restored to 50 requests at 09:54:38." Below is a box where the user can set a limit to the number of tweets to scrape, anad a button that reads "Start scraping"

A screenshot of the dialog which offers the possibility to select the data/metadata the user wants to include in the output file. Each tweet's full text, author's screen_name and URL are selected by default. There is a checkbox to anonymize teets, a dropdown menu to select a file format and a download button. When you choose XLSX as the output format, a checkbox can be ticked to format the worksheet as a table in Excel.

📣 Major update to X-Scraper, which enables users to create corpora of tweets without the need for a subscription to X's unaffordable API: each tweet's full data/metadata is now available to include in the output file (except for TXT files, for now).

fmoncomble.github.io/X-scraper/

#linguistics

29!

How to create a more accessible line chart – Nicola Rennie The default settings for chart software are not guaranteed to be accessible, and often need to be adapted for your own chart. In this blog post, we’ll transform a line chart to make it more accessible...

👀 New blog post! 👀

Ever made a *spaghetti* line chart and wondered how you can make it more accessible and more aesthetically pleasing at the same time? 📈

Read this blog post: nrennie.rbind.io/blog/accessi...

#DataViz #RStats