ChatGPT uses too many dream sequences and Gemini won’t stop describing characters.
AI Fiction Is Easy to Detect Because It's Stupid and Bad, Research Finds
ChatGPT uses too many dream sequences and Gemini won’t stop describing characters.
404media.co
After llms, a lot of times when I m checking someone else code and I see "implementations" from scratch, for example PS matching with caliper using crazy loops instead of MatchIt from @noahgreifer.bsky.social . It makes the work so much harder, because the new func probably has some flaws #rstats
Why?!? Why!? "To adjust for baseline confounding a pseudo-population was generated using propensity score overlap weighting. This created an emulated trial scenario ...." For the description ... it even doesn't account the time varying charact. of vax (GBM - PS) #episky
Come work at Ouroboros University! Where our students submit work produced by AI to be graded by profs using AI. And check out our metrics!! We are highly ranked by AI for all the incredible research papers our profs churn out using AI! H/T @wiringthebrain.bsky.social #IWantNoneOfThis #ResistAI
Richard Sever@richardsever.bsky.social · last mo."the first validated, large-scale framework for judging the quality of life science papers independently of venue, identity, and prestige...Blinded domain-expert professors sided with score three times more often than with journal rank" the-one-percent.qedscience.com 1/n
New interactive blog! "Why Adjusted Regression Coefficients Are Less Descriptive Than They Look" rpsychologist.com/descriptive-...
Epi survey. What is your opinion about "hierarchical models" (not a random effects model) - pubmed.ncbi.nlm.nih.gov/9126524/? It is the idea of building sequential models from "distal to proximal" variables #episky #rstats 1/n
The role of conceptual frameworks in epidemiological analysis: a hierarchical approach - PubMed
Conceptual frameworks provide guidance for the use of multivariate techniques and aid the interpretation of their results in the light of social and biological knowledge.
pubmed.ncbi.nlm.nih.gov
Great article! also the paper: www.sciencedirect.com/science/arti...
Worship me at the office altar: Why narcissistic leaders resist remote work
Leaders have displayed diverging reactions to remote work, with some supporting it and others resisting it. Surprisingly little research has examined …
sciencedirect.com
Kevin Elliott@kjephd.bsky.social · last mo.Read this whole story 🎁: www.nytimes.com/2026/06/22/o...
#statsmeme? (Prompted by a conversation with a student who's double majoring in stats-econ and said that stats courses make more sense to him)
“The truth is that ministers have not the faintest clue what to do”. @gsoh31.bsky.social writes on the quiet collapse of Britain’s universities and why so few people seem to care. www.arguably.uk/p/the-quiet-...
The quiet collapse of British universities
One of our great triumphs and economic advantages is fading away
arguably.uk
Spotted in San Diego. “WE THE STUDENTS DO NOT WANT THIS AI IS MORE HARMFUL THAN HELPFUL”
Taxing the rich: it's good. #mapoli www.raiseupma.org/news/dor-fai...
DOR: Fair Share Revenue Up 20% Over First 10 Months of FY26, Already Exceeding $3 Billion — Raise Up Massachusetts
Tax on Million-Dollar Earners Continues to Exceed Revenue Projections, Proving Fears of Millionaire Flight False BOSTON – The State Department of Revenue yesterday certified that Massachusetts...
raiseupma.org
UC Berkeley Law creates new AI policy that bans AI for "conceptualizing, outlining, drafting, revising, translating, or editing any work submitted for credit" or exams because "thinking remains the sine qua non of good lawyering (and of a quality legal education)." That's what I'm talking about.
I can only suggest to start noticing it as a first step. Time to stop the typical weird and gross habit of the whole world citing very specific USA scholars, when they never even read our work properly let alone cite it, but certainly benefit from it. Honestly, start ignoring these people. 3/
News & Views: New data reveals that a past #tuberculosis diagnosis increases the risk of death up to 14 years later regardless of treatment outcome, which should prompt urgent prioritization of global prevention efforts.
The ghost of tuberculosis past - Nature Medicine
The analysis of the 100 Million Brazilian database reveals that a past tuberculosis diagnosis increases the risk of death up to 14 years later regardless of treatment outcome, which should prompt urgent prioritization of global prevention efforts.
dlvr.it
Looking at a paper, it appears that the Table 2 fallacy (on steroids) was rediscovered by environmental epidemiology. Contrafactual of multiple variables. Where is @cameronpat.bsky.social preaching the word? #rstats #episky
"AI, as it’s being used today, is colliding with institutional incentives to create more rather than better research"
Rebecca Colesworthy@rcolesworthy.bsky.social · 3mo agoAI produces MORE editorial labor, not less. “[To weed out slop] the journal doubled its deputy editors from 6 to 11 & nearly doubled its sr editors from roughly 30 to 60. All of this is volunteer labor, unpaid academics donating time to maintain scientific quality.“ www.forbes.com/sites/johndr...
There are already too many creeps who never spent a second of concern for other people, but that demand rights for algorithms. And it's gonna get so much worse.
LLMs have entered the chat. And the guardrails are off. I've had the "pleasure" of seeing some R code that is clearly the result of some misguided prompt producing some very wrong and worse than useless results. This is bad. It's worse than "decision tree statistics", worse than not knowing. 4/
Why so many "prompt" courses. Ffs... Apart from the normal toxicity from LinkedIn, nlw it is "AI"...
It might seem ridiculous for Nvidia to spend more on AI programming than it would cost to have humans to do the same programming. But it makes sense if you consider Silicon Valley's back-scratch culture, since spending on AI flows to other tech companies, who then funnel it back to Nvidia.
statmodeling.stat.columbia.edu/2026/04/22/f... @adolfont.github.io @olivia.science @irisvanrooij.bsky.social It's strange that in the text, the person states "Without... using it(LLM), you would have had to do the research yourself.... gain understanding... ", but also defendsfend the use of LLM
Fraud and the false optimism of AI for science | Statistical Modeling, Causal Inference, and Social Science
statmodeling.stat.columbia.edu
Waiting @statsepi.bsky.social @deevybee.bsky.social @wiringthebrain.bsky.social to point out possible problems 😀
Nature Medicine@natmed.nature.com · 3mo ago#Microbiome analysis reveals gut microbial changes in #ParkinsonsDisease evolve progressively from healthy individuals to genetically at-risk individuals to #PD patients, with the degree of #dysbiosis correlating with disease progression.
Just read a comment: "... In my use, there is plenty of critical reasoning and direction by me in getting AI to do tasks that I want...." Is it like a gambler addict thinking they can direct the slot machines and they are learning techniques?
In a Brazilian self-controlled case series of 5055 hospitalizations for Guillain–Barré syndrome, laboratory-confirmed dengue was associated with an increased risk of Guillain–Barré syndrome occurring within 42 days. Learn more: nej.md/4cjXU8h
