The fact that AI will give you something that *looks like* an answer is a terrible characteristic of AI. The answers to most questions is not known (especially in medicine) and pretending that the answer is known is just dishonest.

The same applies to medicine. Rich people will have human doctors. Poor people will have LLMs

I am by no means a prominent public intellectual, but my inbox is increasingly filled with messages from people who have been convinced by sycophantic chatbots that they have discovered revolutionary theories that entirely upend our scientific understanding of the universe.

Bellingcat’s contact email has always been a magnet for people with fairly unusual views; paranoid delusions, sprawling conspiracies, the works. But recently, the pattern has shifted, we’re seeing more and more emails clearly written with ChatGPT.

ESICM guidelines on circulatory shock and hemodynamic monitoring 2025. @esicm.bsky.social @intenscaremed.bsky.social #CardioSky Full text: rdcu.be/eQyqN

📖 Respiratory physiology and gas exchange: Blue Book - Chapter 2 🫁 native gas exchange 💨 oxygenation 🩸 O2 content & Hb 🩸 DO2 and VO2 💨 CO2 ⚙️ physiology of #ECLS 🩸 DO2 and CO2 removal in ECLS 📘 In the ELSO #ECMO Specialist Training Manual 5th ed 🛒 bit.ly/ELSObooks

Not now H5N5

This benefit applies for closure with direct suture or with surgical device (Atriclip) as well. The important detail to achieve electrical isolation is the formation of a transmural scar. So amputation can be done, but is optional.

2/ Die nachträgliche Behauptung, es sei um Über- & Fehlbehandlung gegangen, ist durchschaubar. Wer Hochbetagten Therapien streichen will, weil sie angeblich „zu teuer“ sind, spart nicht an Fehlanreizen, sondern an Menschen. Das ist mit Medizin, Ethik & Grundgesetz unvereinbar.

For a surgeon who knows about the variability of LAA anatomy this is no surprise. Even after surgical removal of LAA , there remains a risk of thrombus formation. You do not withdraw ACO. Not all thrombus build up in LAA. The atrium has more folds and crests.

"Don't be evil" is so far in the past, it's not even a memory anymore Google Has Chosen a Side in Trump's Mass Deportation Effort

Alle Schwachen und Wehrlosen, so muss man konstatieren, kommen eher früher als später ins Visier dieser grossen Koalition der Grausamkeit

An AI-powered humanoid robot billed as the first of its kind in Russia fell flat on its face shortly after making its debut appearance at a tech event in Moscow this week.

Genial: die Jobcenter laden psychisch Kranke vor, um zu schauen, ob sie wirklich psychisch krank sind. Wenn sie nicht kommen, sind sie "Totalverweigerer*innen", wenn sie es in die Jobcenter schaffen, können sie ja nicht wirklich psychisch krank sein

Ya don't say 🫩

This little fella did its job very well for nearly half a century (!), but eventually reached the end of its life and had to be replaced by our cardiac surgeons the other day... #cardiosky #ansky #starredwards #ifyouknowyouknow

"An LLM helped me with some task" is not a good argument for these things increasing productivity. Did you also factor in all the cases where it made you slower? Where the output created extra work? And how much of your work/activity is that task?

„Value“ as in money saved, not in hard, patient centred outcomes.

So they have hundreds of millions just lying around that they’ll never miss to give to the government programs immediately, is what they’re announcing

“I already ban the use of generative AI in my own courses, and as a promoter of master’s theses. I explicitly tell my students they won’t learn anything by using it. [...] A general ban is necessary, but nobody dares to say so.” 🙋🏻‍♀️🙋🏻‍♀️🙋🏻‍♀️ apache.be/2025/10/24/b...

gelegentlich zur überprüfung na sichi

“My LLM has a failure rate of 50-80% on the first pass, so I make a second pass to fix the errors.” DAWG, if your LLM fails 50-80% of the time on the first pass, what makes you believe it is failing at less than 50-80% on the second pass? What the fuck.

This sucks so much man wtf

Göbbels hat übrigens auch schon das verdorbene #Stadtbild beklagt, da ging es um Juden*Jüdinnen. Das völkisch-rassistische Denken gleicht sich, auch wenn es die Feindgruppen bisweilen ändert. (Arische) Reinheit & Hegemonie stehen im Zentrum dieser Weltsicht. So denkt leider auch der Bundeskanzler.

so theres this document called a will and trusted people called friends and family

Take a look at the 📦 Equipment section of the ELSO's #ECMO Bedside Guide: 💉 cannulas 💨 membrane lungs ⚙️ pumps 🌡️ temperature regulation This essential, free open access app to support management of patients receiving #ECLS Apple 📱 bit.ly/ELSOBedsideG... Android 📱 bit.ly/ELSOBedsideG...

This looks against every conceivable code of research integrity worthy of the name... what are they thinking?? New depths reached

Using ChatGPT

Just for the benchmark : the historical likelihood of a cancer therapy that enters first human clinical trials (which is way further than this therapy is) to reach FDA approval is about 3.4%. My general advice is not to get too carried away with breakthroughs that treat cancer in mice, sorry!

LLMs are simultaneously magic machines that "democratise" every aspect of intellectual endeavour, able to be used by everyone for all means... But also, LLM use is actually so valid and serious that prompt engineering is a special skill only few cans acquire.

So thankful that dudes without game now have the option of driving all of our energy bills up to come up with devastating lines like it was nice to meet you

@ec.europa.eu please reconsider, read our work: Guest, O., Suarez, M., Müller, B., et al. (2025). Against the Uncritical Adoption of 'AI' Technologies in Academia. Zenodo. doi.org/10.5281/zeno...

Accepting this as an argument means accepting that, as an educator, your primary commitment is not to educating your students, or to propagating knowledge of your subject, but to finding and securing new markets for products in whose success your employers (or their bosses) have some kind of stake.

It is also remarkable how far what is essentially advertising copy has penetrated into ostensibly neutral, scholarly contextualizations even of critical studies — rote invocations of AI’s “power,” “potential,” and ubiquity, untethered to any specific sources or data, are just background noise now.

Auf massive Entrechtung für Menschen in Armut hat sich der Koaltitionsausschuss in der Nacht geeinigt: 1. 100% Sanktion nach 3. verpassten Termin, einschl. Miete 2. 60% bei 2. verpassten Termin 3. 30% Sanktion bei 1. verpassten Termin dunkler Tag heute für @sanktionsfrei.bsky.social

🌟 New preprint 🌟, by @olivia.science and me: 📝 Guest, O., & van Rooij, I. (2025). *Critical Artificial Intelligence Literacy for Psychologists*. doi.org/10.31234/osf... 🧪

Photon-counting detector CT offers high specificity and negative predictive value in detecting in-stent restenosis across stent sizes, highlighting its potential as a reliable non-invasive follow-up tool. #cardiosky #medsky #cardiology t.co/JAWr3gNUHM

Die römischen Kaiser von LinkedIn. Daphne Schwarz (@freihanddenker.bsky.social) schreibt über das Elend der zeitgenössischen Begeisterung für die Stoa, die zur Lieblingsphilosophie der zeitgenössischen Unternehmerkultur aufgestiegen ist. 54books.de/die-roemisch...

🔹According to OCHA estimates, over 200,000 people are estimated to remain in Gaza City, amid ongoing bombardment, deteriorating access to means of survival & a growing number of aid organizations forced to suspend operations. Our latest report on #Gaza & the #WestBank: www.unrwa.org/resources/re...

just sayin‘

British societies guideline on the management of emergencies in patients on extracorporeal membrane oxygenation CCR Journal Watch - tracking the critical care literature daily https://criticalcarereviews.com/latest-evidence/journal-watch

Excellent! Just don't ask it to do math. Or physics. Or chemistry. Or biology. Or astronomy. Or geology. Or geography. Or computer science. Or engineering. Or economics. Or psychology. Or sociology. Or political science. Or anthropology. Or linguistics. Or philosophy. Or art. Or music. Or lit...

I considered writing a long carefully constructed argument laying out the harms and limitations of AI, but instead I wrote about being a hater. Only humans can be haters.

and scene

I know medical people don’t like to be political, but this is happening in a city where an international medical conference will occur in just over two weeks.

An intervention with a NNT of 20 for mortality costs negative dollars, involves taking fewer pills, and has 0 side effects. The intervention is stopping aspirin in patients with chronic CAD (including stents >6 mo old) who take an oral anticoagulant.

It also gets at the psychology of why people get hooked. It's an unpredictable reward structure like gambling. Getting what you want from it is like seeing 🎰

This really gets to the heart of why the economics behind LLM-based genAI do not work. Even if the cost of inference were to come down (it's not), the fact that you have to wrangle the model into a desired output a variable number of times means it will never be efficient enough to be profitable.

What academic worth their salt trusts the word of ANY Big Tech company? Their histories demonstrate they will lie, cheat and steal and that they do not care about anything about making trillions; why trust them? Why entrust our students to them? Why train your chatbot replacements?

now where i am from when someone in your office has an accuracy rate of 57.2 percent in editing spreadsheets we call them "fired"