The Research Desk.

The most upvoted and starred AI research crossing the community today.

Last Brew Time: Oct 3, 2026, 10:55 AM PT

X.com Research Buzz

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians
X.com
15486

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians

Kartik Chandra, Max Kleiman-Weiner, Jonathan Ragan-Kelley, Joshua B. Tenenbaum

MIT CSAIL, University of Washington, Seattle, MIT Department of Brain & Cognitive Sciences

NLP
The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It
X.com
9524

The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It

Valen Tagliabue, Leonard Dung, Cameron Berg

Financing the AI Buildout
X.com
8902

Financing the AI Buildout

Stijn Van Nieuwerburgh

Columbia Business School

What if automating AI R&D triggers an intelligence explosion?
X.com
6509

What if automating AI R&D triggers an intelligence explosion?

Alan Chan, Christoph Winter, Andrew Barto, Jakub Pachocki, Geoffrey Hinton, Eric Horvitz, Yoshua Bengio, Dawn Song, Jack Clark, Hilary Greaves, Anton Korinek, Samuel Hammond, Thore Graepel, Ben Bariach, Philip H. S. Torr, Sheila A. McIlraith, Jeff Clune, Sam Manning, Girish Sastry, Tom Davidson, Daniel Eth, Sören Mindermann

GovAI, CASP, University of Cambridge, University of Massachusetts Amherst, OpenAI, University of Toronto, Vector Institute, Microsoft, Mila (Québec AI Institute), Université de Montréal, LawZero, University of California, Berkeley, Anthropic

AlphaXiv Trending

Towards Looped Models Done Right Part II: Rethinking at Fixed Points
AlphaXiv
128

Towards Looped Models Done Right Part II: Rethinking at Fixed Points

1k, views

Retrieval
ScholarCatalyst: A Benchmark for Retrieving Papers That Inspire New Research
AlphaXiv
85

ScholarCatalyst: A Benchmark for Retrieving Papers That Inspire New Research

3k, views

Affiliation: Stanford University Seoul National University University of Washington, Affiliation: Carnegie Mellon University Allen Institute for AI MIT, Retrieving Papers that Inspire New Research, Stanford University Seoul National University University of Washington, Carnegie Mellon University Allen Institute for AI MIT

NLP
Decoding Looped Transformers Better for (Almost) Free
AlphaXiv
49

Decoding Looped Transformers Better for (Almost) Free

views

Apple

Reinforcement Learning
SkillRefiner: Offline Skill Refinement from Historical Agent Traces
AlphaXiv
45

SkillRefiner: Offline Skill Refinement from Historical Agent Traces

1k, views

Machine Learning
Finetuning with Sampling: SFT Learns Better Than You Think
AlphaXiv
30

Finetuning with Sampling: SFT Learns Better Than You Think

views

Affiliation: Harvard University Website Code, Harvard University

Computer Vision
VISTA: A Visual Harness for Reasoning in an Interactive World
AlphaXiv
23

VISTA: A Visual Harness for Reasoning in an Interactive World

views

Affiliation: Massachusetts Institute of Technology, Affiliation: Massachusetts Institute of Technology Affiliation: Code: https://github.com/joshhhhhan/VISTA, Massachusetts Institute of Technology

HuggingFace Daily Papers

NLP
Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes It
HuggingFace
61

Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes It

Zehao Jin, Ruixuan Deng, Junran Wang

Machine Learning
Video Generation Models: A Survey of Post-Training and Alignment
HuggingFace
43

Video Generation Models: A Survey of Post-Training and Alignment

Chaoyu Li, Xiaoyi Gu, Yogesh Kulkarni, Eun Woo Im, Mohammadmahdi Honarmand

Arizona State University

Persona Dosing: Calibrated Activation Steering for Graded Trait Control
HuggingFace
40

Persona Dosing: Calibrated Activation Steering for Graded Trait Control

Zehao Jin, Junran Wang, Ruixuan Deng, Jiahao Chen, Jingyuan Zhang

Computer Vision
Ego2Act: Evaluating Goal-Directed Manipulation in Egocentric Video Generation
HuggingFace
20

Ego2Act: Evaluating Goal-Directed Manipulation in Egocentric Video Generation

Patrick Amadeus Irawan, Iskandar Muda Rizky Parlambang, Rava Maulana, Qinrong Cui, Erland Hilman Fuadi

Mohamed Bin Zayed University of Artificial Intelligence

Reinforcement Learning
MemFold: Learning Compact Soft Memory for Long-Context Personalization via On-Policy Optimization
HuggingFace
6

MemFold: Learning Compact Soft Memory for Long-Context Personalization via On-Policy Optimization

Jingxuan Wu, Yuzhe Yang, Yiqiao Huang, Chengzhi Liu, Qingni Wang

NLP
Replacing Large Language Models with Jev Decision Models for Low-Latency Edge Service Orchestration
HuggingFace
5

Replacing Large Language Models with Jev Decision Models for Low-Latency Edge Service Orchestration

Delong Li, Xu Wang, Haochen Gong, Rui Lang, Guangsheng Yu