Every week, I organize 400 to 700 links into roughly 60 categories as part of my ongoing effort to learn about AI. This is my personal notebook, which I enjoy sharing with friends… a hobby and a labor of love, rather than a commercial publication or product.
If you arrived here through a search or shared link, this page collects the links I found for Science and Medicine for the week ending August 28, 2026.
As part of my learning process, I like to automate the category covers. It gives me a chance to learn Python and APIs.
This week’s cover prompt was written using Claude Opus 4.7, and the image was generated using Nano Banana 2.1.
Category Cover Image Prompt: A 16:9 mid-1980s murder-mystery movie poster of a broad three-story Victorian mansion in dense black-and-white pen-and-ink engraving on pure black, with arched windows glowing warm amber showing costumed science suspects (a lab-coated researcher with a bubbling flask, a statistician holding a chart, a goggled chemist mid-eureka), still-life windows crowded with DNA helices, petri dishes, a centrifuge, lab notebooks and a brain in a jar, plus one window with a startled lab rat silhouette, a silver-haired mustachioed butler in black tailcoat standing in the open front door holding a silver tray with a sealed envelope marked with a tiny atom, and the fat rounded white title Science above the roofline. Symmetrical frontal composition, crosshatched engraving, campy spooky theatrical lighting, no extra text.
This Week in Science and Medicine News
Summary by Claude Sonnet 5.5 based on this week’s links:
- Agents still struggle with real science work: Two new benchmarks found that the best agents fall short on full scientific tasks. Bixbench3 had agents reproduce papers end-to-end and reported frontier agents scoring below 50%, with problems like quitting or faking data. SWE-bench Science reported that even Claude Code with Opus-5 passes under 50% of its tasks.
- Clinical AI is judged on reasoning, not just answers: Scale AI Labs introduced CliniCARE-Bench, arguing that clinical agents can be right for the wrong reasons. It tests whether they can work through patient records, ground conclusions, and know when to abstain.
- AI moves into real medicine and forecasting: University College London Hospitals NHS reported a world-first AI-assisted brain surgery that saved a patient's sight. Google also pointed to WeatherNext, its AI model for forecasting cyclones, and Google Research introduced GlucoFM for glucose monitoring data.
This summary was generated by Claude Sonnet 5.5 to help you explore the links below. Rest assured, I select, organize, and check the links by hand in Google Sheets, and write the introduction and personal commentary in The Main Newsletters myself each week as a labor of love.
This week's links related to Science and Medicine
When people ask: “what personal tasks do I need an agent for?” I think that it is because they think of automation as a simple regular task for the AI. But frontier AI is now really good at helping with irregular, draining tasks: a medical bill, a credit dispute, tedious forms…”
https://x.com/emollick/status/2091893964311855222
Starting today, 10,000 scientists across every field, from math to chemistry to physics and more, can get Claude through our new Claude Team plan for scientists. Standard seats are free, and premium seats with 5x usage limits are $15 per month, an 80% discount, for one year.”
https://x.com/claudeai/status/2093059087298601113
WeatherNext: Our AI model helping to forecast cyclones”
https://x.com/GoogleAI/status/2092275116503707733
World-first AI-assisted brain surgery saves patient’s sight
https://www.uclh.nhs.uk/news/first-patient-live-ai-assisted-sight-saving-brain-surgery
SWE-bench Science A new benchmark for scientific software engineering: 119 tasks across 98 repositories and 20 domains. Even the best agent, Claude Code with Opus-5, achieves under 50% pass@1.”
https://x.com/HuggingPapers/status/2090773411039457342
GPT-5.6 Sol seems to currently be the most capable model available for almost every single task (at least the ones I am testing), across coding, math, agentic workflows, prose and creative writing. It’s likely going to go down as the Opus 4.6 of OAI, and a much more significant”
https://x.com/DimitrisPapail/status/2090589493984465321
Today, we’re kicking off the first phase of the research preview for Model Hardware Standard (MHS): a new standard for AI agents to safely operate physical equipment in scientific research and advanced manufacturing. Read more:
https://x.com/AnthropicAI/status/2093038426140651791
Fable: “I want a twigl shader that renders Lost Carcosa from The King in Yellow.” (Shaders are procedurally generated using math alone)
https://t.co/IBRjbPe96d Along the shore the cloud waves break, The twin suns sink behind the lake, The shadows lengthen In Carcosa”
https://x.com/emollick/status/2091001394534707474
Bixbench3: Can models reproduce the entire analysis from papers? We had agents repro papers end-to-end, with over 1B tokens spent on some. Frontier agents still score below 50%, and due to real problems like environment misconfig, quitting, or faking data 1/6″
https://x.com/andrewwhite01/status/2092650535119900968
Clinical AI agents can be right for the wrong reasons. Real clinical work requires more than just medical knowledge. Agents need to navigate longitudinal records, reconcile evidence, ground conclusions, and know when to abstain. We introduce CliniCARE-Bench, a new benchmark”
https://x.com/ScaleAILabs/status/2092695734852476957
[2408.06292] The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
https://arxiv.org/abs/2408.06292
Math is fine, but vague posting from labs about new models should only be allowed if the post consists solely of the output from the prompt in the Sparks of AGI paper: “Draw a unicorn in TiKZ.
https://x.com/emollick/status/2091995606516441405
Always bring the donuts. And the protein shakes.”
https://x.com/bilawalsidhu/status/2091182982727946384
Large language models are trained on millions of authors … but converge on a single, homogeneous voice. A group of psychologists and computer scientists developed PsychAdapter to reintroduce human variation into AI-generated text, opening new possibilities for therapy training”
https://x.com/StanfordHAI/status/2092028830471119010
Terminal-Bench-Science 0.1
https://www.terminal-bench-science.ai/announcement
The University of Michigan put their entire robotics degree on GitHub. Not one course. The whole curriculum. ROB 101 … Computational Linear Algebra for Robotics ROB 311 … How to Build Robots and Make Them Move ROB 501 … Mathematics for Robotics ROB 530 … Mobile Robotics Every”
https://x.com/IlirAliu_/status/2092883820877926851
Introducing GlucoFM, a lightweight, self-supervised continuous glucose monitoring foundation model that separates metabolic baselines from transient spikes, producing transferable representations and setting new performance standards across diverse metabolic prediction tasks,”
https://x.com/GoogleResearch/status/2092686085910667663
We use Inkling-Small to turn paper abstracts into quick, useful summaries. open weights × open science 🤝”
https://x.com/huggingface/status/2092727360584135026
PSA: for everyone using GLM-5.3 Flash just use “high” reasoning_effort unless you’re asking it to curing cancer or something > “high” 28% acc, 70k avg tokens > “max” 28% acc, 140k avg tokens”
https://x.com/zainhas/status/2093125213361938621
Scaling Muon for Diffusion Transformers New paper by Meta and USC. “We first establish Muon’s scaling behavior on DiTs from 1.3B to 15B parameters, showing that its optimization and generative quality advantages over AdamW persist across model scales. However, at scale, the”
https://x.com/iScienceLuvr/status/2091820249226293576
How is AI transforming empirical research? HAI’s Stanford Causal Science Center and @StanfordGSB are hosting the Empirical Methods in the Age of AI Conference on October 2-3, 2026 … bringing together leading academics and industry practitioners to explore that question across”
https://x.com/StanfordHAI/status/2092284099193147518
Michael Polansky is training an AI model on skin that’s still alive | TechCrunch
https://techcrunch.com/2026/08/21/michael-polansky-is-training-an-ai-model-on-skin-thats-still-alive/
Stanford HAI’s AI Index is hiring a Research Scientist to monitor emerging research, collect and validate data, and help measure AI’s trajectory for one of the most widely cited resources in the field. Apply here:
https://x.com/StanfordHAI/status/2093007676909420761
Time to fake a deep knowledge of mathematics so I can have an opinion of how impressive this is.”
https://x.com/emollick/status/2091734781935456365
Why we are building Outer Bio | Outer Bio
https://www.outerbio.com/news/why-we-are-building-outer-bio
📈 New blog post: Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers. As a practical example, I finetuned a ColBERT-style model for medical retrieval. 14.5 hours on one RTX 3090, and it beats every general-purpose retriever I could find. Thread 🧵”
https://x.com/tomaarsen/status/2092611931890713066
Leave a Reply