Hacker News with Generative AI: Research

Germany: Far-right political bias ahead of federal elections in X and TikTok (techcrunch.com)
Recommendation algorithms operated by social media giants TikTok and X have shown evidence of substantial far-right political bias in Germany ahead of a federal election that takes place Sunday, according to new research carried out by Global Witness.
Do ambiguous images provide psychological insights? Testing a popular claim (peerj.com)
Social media posts and websites claim that the way in which people perceive ambiguous images reveals insights into their personality and thinking style.
After 20 Years, Math Couple Solves Major Group Theory Problem (quantamagazine.org)
In 2003, a German graduate student named Britta Späth encountered the McKay conjecture, one of the biggest open problems in the mathematical realm known as group theory.
Show HN: ArXiv-txt, LLM-friendly ArXiv papers (arxiv-txt.org)
Make arXiv papers easily available in LLM-friendly formats.
What Makes a Great Software Engineer (Dissertation) (2016) [pdf] (faculty.washington.edu)
Muse: Our first generative AI model designed for gameplay ideation (microsoft.com)
Today, the journal Nature (opens in new tab) is publishing our latest research, which introduces the first World and Human Action Model (WHAM). The WHAM, which we’ve named “Muse,” is a generative AI model of a video game that can generate game visuals, controller actions, or both.
Accelerating scientific breakthroughs with an AI co-scientist (research.google)
We introduce AI co-scientist, a multi-agent AI system built with Gemini 2.0 as a virtual scientific collaborator to help scientists generate novel hypotheses and research proposals, and to accelerate the clock speed of scientific and biomedical discoveries.
Broken legs and ankles heal better if you walk on them within weeks (scientificamerican.com)
Twenty years ago my husband, Mark, broke his left ankle and was in a cast and on crutches for nearly two months. Last year he broke the other ankle. But this time, after surgery, his doctor surprised us by instructing Mark to walk on it two weeks later.
Older AI models show signs of cognitive decline, study shows (livescience.com)
Ask HN: What are the most important AI papers/ online resources of 2024/2025? (ycombinator.com)
Ask HN: What are the most important AI papers/ online resources of 2024/2025?
Thinking Machines Lab (thinkingmachines.ai)
Thinking Machines Lab is an artificial intelligence research and product company. We're building a future where everyone has access to the knowledge and tools to make AI work for their unique needs and goals.
Among top researchers 10% publish at unrealistic levels, analysis finds (chemistryworld.com)
About 10% of the most influential researchers worldwide in various scientific fields, including chemistry, are achieving ‘implausibly high’ publication and new co-author rates.
X Prevents Research on Potential Election Interference (freiheitsrechte.org)
The social media platform X denied researchers access to public data on its platform. Together with Democracy Reporting International (DRI), we achieved a major success in summary proceedings.
Census reveals high disability rates among gender-diverse Canadians (medicalxpress.com)
Using data from the latest census, York University researchers from the Faculty of Health found Canadians who identify as gender diverse experience disability at rates much higher than their cis counterparts. In particular, nonbinary individuals consistently had the highest levels of disabilities, followed by binary-transgender individuals.
School phone bans do not improve grades or wellbeing, says UK study (theguardian.com)
Banning smartphones at school does not by itself improve academic grades and children’s wellbeing, a study suggests.
EnigmaEval: A Benchmark of Long Multimodal Reasoning Challenges (arxiv.org)
As language models master existing reasoning benchmarks, we need new challenges to evaluate their cognitive frontiers.
Large Language Models Show Concerning Tendency to Flatter Users (xyzlabs.substack.com)
Recent research from Stanford University has revealed a concerning trend among leading AI language models: they exhibit a strong tendency toward sycophancy, or excessive flattery, with Google's Gemini showing the highest rate of such behavior.
Alzheimer's Disease as Type 3 Diabetes (gethealthspan.com)
Over the past several decades, Alzheimer’s disease (AD) has been predominantly explained through the lens of amyloid-beta plaques and neurofibrillary tangles, yet an emerging paradigm shift increasingly recognizes metabolic dysfunction—specifically, insulin resistance—as a key driver of AD.
Carbon capture more costly than switching to renewables, researchers find (techxplore.com)
For most countries around the world, sourcing energy entirely from wind, solar, geothermal, and hydropower by 2050 would reduce their energy needs and costs, improve air quality, and help slow climate change, according to a study in Environmental Science & Technology.
Can We Trust AI Benchmarks? A Review of Current Issues in AI Evaluation (arxiv.org)
Quantitative Artificial Intelligence (AI) Benchmarks have emerged as fundamental tools for evaluating the performance, capability, and safety of AI models and systems.
The Impact of Generative AI on Critical Thinking [pdf] (microsoft.com)
Researchers from Microsoft and Carnegie Mellon University warn that the more you use AI, the more your cognitive abilities deteriorate.
The danger of relying on OpenAI's Deep Research (economist.com)
Economists are in raptures, but they should be careful
In a showdown of psychotherapists vs. ChatGPT, the latter wins, new study finds (fortune.com)
Couples therapy can be helpful—but so can an AI chatbot, researchers found.
Barcoding brains (asimov.press)
Connectomics — a technique that maps physical connections between neural cells — is expensive and inefficient. E11 Bio, a non-profit research group, is designing a tool to expedite progress.
Ask HN: Seeking Book (ycombinator.com)
Hello All,<p>Many years ago I came across a book from a female Researcher which was basically using low-level mathematics to model low-level type 'circuits' or 'functions'. Not circuits in the electrical sense but I do know this researcher eventually did work for some Intelligence Agencies as part of extensions to her work. Problem I have is I have searched high and low via various search engines though with proliferation of AI texts it has increased the noise level.
A study on how turtles navigate using the Earth’s magnetic field (unc.edu)
Carolina researchers publish a groundbreaking study on how turtles navigate using the Earth’s magnetic field.
Show HN: A New Way to Learn Languages (langturbo.com)
Learn faster with podcasts and the latest language learning research.
Does X cause Y? An in-depth evidence review (2021) (cold-takes.com)
There's an interesting theory out there that X causes Y. If this were true, it would be pretty important. So I did a deep-dive into the academic literature on whether X causes Y. Here's what I found.
How a computer that 'drunk dials' videos is exposing YouTube's secrets (bbc.com)
YouTube is about to turn 20. An unusual research method is unveiling statistics about the platform that Google doesn't want you to know.
UTEP leaders silent as Ted Cruz accuses researchers of using 'woke DEI grants' (elpasomatters.org)
UTEP researchers won a prestigious $1 million grant in 2022 to better understand the evolution of birds. U.S. Sen. Ted Cruz, R-Texas, says it’s one of thousands of National Science Foundation grants in recent years – including 17 to the University of Texas at El Paso – that are examples of “how the Biden administration weaponized federal agencies to push a far-left ideology.”