Links from June 2025: changing minds, Alzheimer’s drugs; copyright in favour of AI
Predicting genome differences; image hosting; experience & relationships beat reason; Alzheimer’s drugs; copyright judge rules in favour of AI
Adding images to posts on the Pagecord free plan
Pagecord
Pagecord is a blogging platform I’m tracking, not yet using. They’ve done the hard work of researching image hosting: PostImage, cubeupload, ImageKit.
Anthropic wins a major fair use victory for AI — but it’s still in trouble for stealing books
The Verge
A case in US law: “The judge ruled that digitizing a legally purchased physical book was fair use, and that using those digital copies to train an LLM was sufficiently transformative to also be fair use.” Seen via Simon Willison’s blog.
This article won’t change your mind. Here’s why
The Guardian
There are lots of reasons why debate (and indeed, information-giving and argumentation in general) tends to be ineffective at changing people’s political beliefs […] probably the most important reason words don’t change minds is that two other factors carry far more influence: our social relationships; and our own actions and experiences.” Via @orbific.
A Patient-in-the-Loop Approach to Artificial Intelligence in Medicine
JAMA Network Open invited commentary
This is a summary of a longer paper which is a “study surveying 13 806 patients […] from 74 hospitals in 43 countries found that while patients were generally supportive of AI-enabled health care facilities and recognized the potential of AI, they preferred explainable AI systems and physician-led decision-making.”
Britain’s AI-care revolution isn’t flashy—but it is the future
The Economist
Close to a profile of Cera. “With its app, which uses algorithms to predict fall risk, Cera claims to have cut falls by a fifth. A peer-reviewed study from 2022 found that its app had reduced hospitalisations by 52%. In March the NHS said it would work with Cera to roll out its AI tool across the country.”
Neurosymbolic AI is the answer to large language models’ inability to stop hallucinating
The Conversation
“The reality is that if AI is going to keep advancing, we will need systems that adapt to novelty from only a few examples, that check their understanding, that can multitask and reuse knowledge to improve data efficiency and that can reason reliably in sophisticated ways.”
Roger Penrose – Why Intelligence Is Not a Computational Process
Breakthrough Discuss 2025, YouTube
My AI indoctrination says: machines will be able to think, because we view the
mind as a machine, and minds think. So I’m interested in why some say otherwise. Around 19 minutes into this video the argument goes: 1) Physics doesn’t have the laws for the physical process of wave function collapse; 2) RP claims (a) “understanding” (and also consciousness) is something to do with wave function collapse, and (b) collapse is not computational; 3) Therefore, machines will not “understand”. There are lots of leaps there I don’t follow.
Quality and Outcomes Framework guidance for 2025/26
NHS England
GPs earn points for targets, such as blood pressure monitoring, or smoking cessation. Points convert to cash at some point. QOFs are pronounced “quoff”
The Alzheimer’s drug pipeline is healthier than you might think
The Economist
"There are 182 clinical trials for Alzheimer’s treatments under way in 2025—an 11% increase on the previous year—testing 138 different drugs, of which 12 are set to complete their final “phase 3” trials this year. Moreover this pipeline includes medicines aimed at a diverse range of targets in the brain, reflecting an increasingly sophisticated understanding of the molecular processes behind Alzheimer’s and dementia more broadly.”
Google’s new AI will help researchers understand how our genes work
MIT Technology Review
“AlphaGenome, an AI model that predicts what effects small changes in DNA will have on an array of molecular processes, such as whether a gene’s activity will go up or down.”