Newest Papers
How “Cozy Lit” Became the Latest and Most Shameless Form of Digital Escapism | The Walrus
Greta Rainbow
The Walrus·2025
Think cats. Tea. Rain. The seaside. More cats
No ratings yet
View paper →How “The New Yorker at 100” Got to Netflix | The New Yorker
Jelani Cobb
The New Yorker·2025
The documentary, available to stream on December 5th and featuring Sarah Jessica Parker, Jon Hamm, Julianne Moore, and staff writers and editors, is discussed by the filmmakers Judd Apatow and Marshall Curry with Jelani Cobb.
No ratings yet
View paper →AI & Human Co-Improvement for Safer Co-Superintelligence
Jason Weston, Jakob Foerster
arXiv·2025
Self-improvement is a goal currently exciting the field of AI, but is fraught with danger, and may take time to fully achieve. We advocate that a more achievable and better goal for humanity is to maximize co-improvement: collaboration between human researchers and AIs to achieve co-superintelligence. That is, specifically targeting improving AI systems' ability to work with human researchers to conduct AI research together, from ideation to experimentation, in order to both accelerate AI research and to generally endow both AIs and humans with safer superintelligence through their symbiosis. Focusing on including human research improvement in the loop will both get us there faster, and more safely.
No ratings yet
View paper →Persistent Attenuation of Lymphocyte Subsets After Mass SARS-CoV-2 Infection
Zhengqi Jiang, Tichao Shan, Yucan Li, Fengjiao Han, Baobao Feng, Xiaohui Zhen, Heyu Ni, Jun Peng, Miao Xu
sciencedirect.com·2025
Objectives
Growing evidence suggests that lymphocyte subsets are declined in COVID-19 patients, but it is unclear if these alterations persist after widespread exposure to SARS-CoV-2 or how long they last.
Methods
We analyzed lymphocyte subset data from 40,537 patients across three phases: pre-COVID, mass infection, and post-COVID. The counts of lymphocyte subsets and CD4+/CD8+ ratios were compared using Mann-Whitney U test or Kruskal-Wallis H test. Monthly post-exposure data were compared with pre-exposure data to assess the persistence of impact on lymphocyte subsets by SARS-CoV-2, and subgroup analyses were performed in patients with cardiovascular disease.
Results
During mass infection, T cells, CD4+T cells, CD8+T cells, NK cells, and B cells dropped significantly. Even 20 months post-infection, CD8+ T cells remained 9.9% below baseline. Baseline lymphocyte subsets differed significantly by sex and age. Immune recovery varied by age and sex, with older adults and males showing prolonged lymphopenia. In cardiovascular disease patients, T lymphocytes remained 72.9% below baseline for 20 months post-infection.
Conclusions
Our findings redefine SARS-CoV-2 infection as a condition of long-lasting immune compromise. The sustained subnormal lymphocytes—particularly in cardiovascular disease cohorts—highlight a key immunologic feature of long COVID and underscore the need for personalized care.
No ratings yet
View paper →Molecular basis for de novo thymus regeneration in a vertebrate, the axolotl
Anna Czarkwiani, Macrina Lobo, Lizbeth Airais Bolaños Castro, Andreas Petzold, Fabian Rost, René Maehr, Maximina H. Yun
Science Immunology·2025
<jats:p>
The thymus is the primary site of T cell development, central to the establishment of self-tolerance and adaptive immune function. In mammals, the thymus undergoes age-related involution, resulting in a global decline in immune function. The thymus has some regenerative ability that relies on pre-existing thymic remnants but is insufficient to prevent involution. Here, we show that the juvenile axolotl (
<jats:italic toggle="yes">Ambystoma mexicanum</jats:italic>
) is able to regenerate its thymus de novo after complete removal, constituting an exception among vertebrates. Using single-cell transcriptomics and genetic and transplantation approaches, we demonstrate that de novo thymus regeneration results in the restoration of morphology, cell-type diversity, and function. FOXN1, although it has a conserved role in thymus organogenesis, is dispensable for the initiation of thymic regeneration. In contrast, we identify midkine signaling as a possible early driver of de novo thymus regeneration. This study demonstrates an instance of organ-level regeneration of the lymphoid system, which could guide future clinical strategies seeking to promote thymus regrowth.
</jats:p>
No ratings yet
View paper →Gender-affirming health care needs, barriers to care, and health and wellbeing in a broad nationwide sample of transgender people in Norway
Silje-Håvard Bolstad, Norman Anderssen, Bjørge H. Hansen, Ilan H. Meyer
BMC Public Health·2025
<jats:title>Abstract</jats:title>
<jats:sec>
<jats:title>Background</jats:title>
<jats:p>Transgender people report increased prevalence of mental distress and suicide attempts compared to the general population. Gender-based minority stressors such as discrimination and victimization are positively associated with mental health problems while social and medical gender transition is inversely associated with mental health problems. Barriers to accessing gender-affirming health care is reported by transgender people in many countries, including Norway which has a state-funded public health care system. The aim of the current study was to examine gender-affirming health care needs, barriers to care and health and wellbeing in a broad nationwide sample of transgender people in Norway.</jats:p>
</jats:sec>
<jats:sec>
<jats:title>Methods</jats:title>
<jats:p>A nationwide sample with 579 transgender participants completed an anonymous online survey during June-September 2023. Chi-square tests and ANOVA were used to examine differences between gender identity groups.</jats:p>
</jats:sec>
<jats:sec>
<jats:title>Results</jats:title>
<jats:p>Having obtained gender-affirming health care was more common among trans men and trans women as compared to nonbinary people. Participants generally reported being satisfied with the outcomes of gender-affirming hormonal and surgical treatments. Having obtained treatment entirely through private funding was reported by 32.5% of those using hormones and 49.5% of those who had obtained surgery. Not being able to afford hormones or surgery was reported by 45.9% and 65.5% among those with unmet treatment needs. Compared to trans men and trans women with unmet needs, nonbinary people with unmet needs were less likely to be under assessment or on a waiting list to obtain treatments. Of the total sample, 36.7% reported suicide attempts, 74.8% reported mental distress above clinical cut-off and 12.5% reported being satisfied with life.</jats:p>
</jats:sec>
<jats:sec>
<jats:title>Conclusions</jats:title>
<jats:p>Transgender people in Norway reported high levels of mental distress and suicidality. Despite being young and with low income, a large proportion had obtained gender-affirming medical treatments entirely through private funding. Transgender people in need of gender-affirming health care could benefit from increased access to care through the state-funded public health care system in Norway.</jats:p>
</jats:sec>
No ratings yet
View paper →Pluralistic: The Reverse-Centaur’s Guide to Criticizing AI (05 Dec 2025)
Cory Doctorow
t.co·2025
Over the summer I wrote a book about what I think about AI, which is really about what I think about AI criticism, and more specifically, how to be a good AI critic. By which I mean: "How to be a critic whose criticism inflicts maximum damage on the parts of AI that are doing the most harm." I titled the book The Reverse Centaur's Guide to Life After AI, and Farrar, Straus and Giroux will publish it in June, 2026.
But you don't have to wait until then because I am going to break down the entire book's thesis for you tonight, over the next 40 minutes. I am going to talk fast.
No ratings yet
View paper →Pluralistic: The Reverse-Centaur’s Guide to Criticizing AI (05 Dec 2025) – Pluralistic: Daily links from Cory Doctorow
Cory Doctorow
pluralistic.net·2025
No abstract available
★ 4.0 (1)
View paper →Expanding Our View of Drosophila Centrioles
Emma Burns, Anastasia Amoiroglou, Carey J. Fagerstrom, John M. Ryniawec, LingSze Lee, Rose K. Runyan, Leah F. Rosin, Gregory C. Rogers, Nasser Rusan
Journal of Cell Science·2025
A significant challenge in Drosophila centriole biology is its small size. Advanced super-resolution techniques have provided valuable insights, but require specialized equipment and can be difficult to implement in tissues. Expansion Microscopy (ExM) offers an accessible alternative, yet its application in Drosophila centriole research has been sparse. We provide an ExM protocol for cultured S2 cells and fly tissues that revealed new insights into pro-centriole biology. In S2 cells we document overduplication in the form of the classic "rosettes", while in spermatids we uncover an unexpected movement of the pro-centriole-like structure (PCL). ExM has also refined existing molecular models. In S2 cells we document the distal tip protein Cep97 as a ring, which clarifies its role in capping the growing centriole. In spermatids, we spatially segregated the inner nuclear membrane protein Spag4 and the cytoplasmic protein Yuri, which led to the new hypothesis that they play independent roles at the centriole-nucleus contact site. Finally, we show that our ExM protocol is a hypothesis-generator and discovery tool applicable beyond Drosophila centrioles by imaging synaptonemal complexes in the Plodia interpunctella moth.
No ratings yet
View paper →The Reckoning We Still Refuse
Future Tense
Substack·2025
It was the second year of his soiree, which drew the 1 percent of the 1 percent of the 1 percent for a Labor Day bash that was described as the closest thing to a night at Studio 54 or the Playboy Mansion.
For over a decade, Puff’s White Party was the flagship event of his social calendar. It was his calling card. A major part of the mythmaking that took him from scrappy label intern to young A&R executive to hotshot producer to millionaire rap mogul to global bon vivant. A Black Jay Gatsby if you will.
★ 4.0 (1)
View paper →The Universal Weight Subspace Hypothesis
Prakhar Kaushik, Shravan Chaudhari, Ankit Vaidya, Rama Chellappa, Alan Yuille
arXiv·2025
We show that deep neural networks trained across diverse tasks exhibit remarkably similar low-dimensional parametric subspaces. We provide the first large-scale empirical evidence that demonstrates that neural networks systematically converge to shared spectral subspaces regardless of initialization, task, or domain. Through mode-wise spectral analysis of over 1100 models - including 500 Mistral-7B LoRAs, 500 Vision Transformers, and 50 LLaMA-8B models - we identify universal subspaces capturing majority variance in just a few principal directions. By applying spectral decomposition techniques to the weight matrices of various architectures trained on a wide range of tasks and datasets, we identify sparse, joint subspaces that are consistently exploited, within shared architectures across diverse tasks and datasets. Our findings offer new insights into the intrinsic organization of information within deep networks and raise important questions about the possibility of discovering these universal subspaces without the need for extensive data and computational resources. Furthermore, this inherent structure has significant implications for model reusability, multi-task learning, model merging, and the development of training and inference-efficient algorithms, potentially reducing the carbon footprint of large-scale neural models.
No ratings yet
View paper →The behavioral selection model for predicting AI motivations
Alex Mallen, Buck
AI Alignment Forum·2025
Highly capable AI systems might end up deciding the future. Understanding what will drive those decisions is therefore one of the most important questions we can ask.
Many people have proposed different answers. Some predict that powerful AIs will learn to intrinsically pursue reward. Others respond by saying reward is not the optimization target, and instead reward “chisels” a combination of context-dependent cognitive patterns into the AI. Some argue that powerful AIs might end up with an almost arbitrary long-term goal.
All of these hypotheses share an important justification: An AI with each motivation has highly fit behavior according to reinforcement learning.
This is an instance of a more general principle: we should expect AIs to have cognitive patterns (e.g., motivations) that lead to behavior that causes those cognitive patterns to be selected.
In this post I’ll spell out what this more general principle means and why it’s helpful. Specifically:
* I’ll introduce the “behavioral selection model,” which is centered on this principle and unifies the basic arguments about AI motivations in a big causal graph.
* I’ll discuss the basic implications for AI motivations.
* And then I’ll discuss some important extensions and omissions of the behavioral selection model.
This post is mostly a repackaging of existing ideas (e.g., here, here, and here). Buck provided helpful discussion throughout the writing process and the behavioral selection model is based on a cau...
No ratings yet
View paper →