Scott McGrath

Alaska Ignored Warning Signs of a Budget Crisis. Now It Doesn’t Have Funding to Fix Crumbling Schools. Lawmakers only budgeted $40 million of the nearly $800 million that districts say is needed to fix and maintain schools to keep them safe and operating. Gov. Mike Dunleavy then vetoed more than…

Alaska's reliance on oil revenue has led to a predictable budget crisis. With lawmakers funding only a fraction of the $800 million needed for school repairs, and most of that vetoed, rural schools continue to crumble, posing risks to students.

What Do Digital Health Leaders Think of Trump’s New AI Action Plan? - MedCity News In a new action plan, the White House aims to boost the country’s AI innovation through deregulation, infrastructure expansion and new incentives. Overall, the move is welcomed by healthcare AI…

Healthcare leaders view the new White House AI plan with cautious optimism. While its pro-innovation stance is welcomed, significant omissions like AI safety, patient consent, and the role of key health regulators raise serious concerns.
#MedSky #MLSky

If you know a high school student who is interested in medicine, computer science, and/or AI, I highly recommend you encourage them to apply for this program.

The deadline is Sept. 9th!

Link: amia.secure-platform.com/symposium/pa...

#MedSky #MedAI

AMA and other medical associations are kicked out of CDC vaccine workgroups U.S. health officials have told more than a half-dozen of the nation’s top medical organizations that they will no longer help establish vaccination recommendations.

U.S. health officials have excluded major medical groups, including the AMA and AAP, from the workgroups that form vaccine recommendations. Citing "bias", the move has been called dangerous by the groups, who warn it will undermine public trust.
#MedSky 🛟

How Trump Is Transforming the U.S. Government’s Environmental Role The E.P.A. said this week it would revoke its own ability to fight climate change. It’s the latest move in an extraordinary pivot away from science-based protections.

The EPA is moving to revoke its 2009 endangerment finding, the scientific basis for regulating greenhouse gases. This action represents a fundamental pivot from decades of science-based protection, shifting climate risk from the state to the individual.

Ukraine rescues soldier via drone delivery of complete e-bike Drones can now carry significant payloads.

Ukraine rescues soldier via drone delivery of complete e-bike

arstechnica.com/gadgets/2025...

Persona vectors: Monitoring and controlling character traits in language models A paper from Anthropic describing persona vectors and their applications to monitoring and controlling model behavior

#MLSky Direct link to the paper: www.anthropic.com/research/per...

Forcing LLMs to be evil during training can make them nicer in the long run New Anthropic research shows that undesirable LLM traits can be detected—and even prevented—by examining and manipulating the model’s inner workings.

A new study offers a counterintuitive method for LLM safety. By artificially activating an "evil" neural pattern during training, researchers found they could prevent the model from learning that behavior from flawed data, without impairing its performance. #MLSky

Quantifying large language model usage in scientific papers - Nature Human Behaviour Liang et al. estimate the prevalence of text modified by large language models in recent scientific papers and preprints, finding widespread use (up to 17.5% of papers in computer science).

🧪 #MLSky #AcademicSky Direct link to the paper: www.nature.com/articles/s41...

One-fifth of computer science papers may include AI content A surge in AI-generated text has been detected in papers across many disciplines since the release of ChatGPT

🧪 A new study of over one million papers reveals a sharp rise in AI-generated content in scientific writing since late 2022. The trend is most pronounced in computer science, where up to 22% of recent papers show signs of LLM modification.
#MLSky #AcademicSky

This company claimed to ‘de-extinct’ dire wolves. Then the fighting started. Colossal’s bold announcements have drawn criticism from many scientists, but the billion-dollar firm is not backing down.

🧪 Colossal Biosciences' announcement of a 'de-extinct' dire wolf, created via gene editing, has sparked a chasm in the scientific community. The debate centers on whether these animals are true resurrections or proxies, igniting conflict over definitions and ethics.
🧬💻

Who’s to blame when AI agents mess up? We urgently need a new system of ethics The deployment of capable AI agents raises fresh questions about safety, human–machine relationships and social coordination.

As AI evolves from tools to autonomous agents, our ethical frameworks are becoming inadequate. There is already the hanging questions of liability, goal alignment, and the nature of human-AI social bonds. We are going to need a new system of evaluation and governance.
#MLSky

Hundreds of Medieval Medical Manuscripts with Strange Cures Get Digitized & Put Online: From Leeches to Crushed Weasel Testicles If any discussion of medieval medicine gets going, it's only a matter of time before someone brings up leeches.

Medieval medical manuscripts, including bizarre remedies like leeches and crushed weasel testicles, have been digitized and made available online, offering a chance to look back into ancient medical practices and beliefs.

Inside OpenAI’s quest to make AI do anything for you | TechCrunch OpenAI developed the first AI reasoning model less than a year ago, but the technology has shifted Silicon Valley's focus to agents.

Inside OpenAI’s quest to make AI do anything for you

techcrunch.com/2025/08/03/i...

Trump Went to War With the Ivies. Community Colleges Are Being Hit.

the trump administration is selling its war on higher education as a strike against the elites who have disadvantaged ordinary americans. in actuality, the administration is destroying advanced education for *everyone* in pursuit of its effort to deskill the american public

What’s Inside the Tiny Miracle Food Pouches That Can Save the Lives of Starving Gazans Packed with calories and protein, the same magic mixture has successfully treated famine for decades—but due to funding cuts it's now in short supply.

What’s Inside the Tiny Miracle Food Pouches That Can Save the Lives of Starving Gazans

www.wired.com/story/whats-...

Reporting guideline for chatbot health advice studies: the Chatbot Assessment Reporting Tool (CHART) statement The Chatbot Assessment Reporting Tool (CHART) is a reporting guideline developed to provide reporting recommendations for studies evaluating the performance of generative artificial intelligence (AI)-...

Reporting guideline for chatbot health advice studies: the Chatbot Assessment Reporting Tool (CHART) statement bmjmedicine.bmj.com/content/4/1/... #llms #chatbots #bioinformatics #healthcare

A Hiker Was Missing for Nearly a Year—Until an AI System Recognized His Helmet Using AI to analyze thousands of frames taken by drone, The Piedmontese Alpine Rescue team has found the body of a doctor who had been missing since September 2024.

A Hiker Was Missing for Nearly a Year—Until an AI System Recognized His Helmet

www.wired.com/story/missin...

A photo of a yellow street sign with the outline of a deer as a warning for deer crossings. The deer outline has a medium sized red sticker placed where the deer's nose would be. The sign has a blue sky behind it with some small whispy clouds in the lower right corner of the photo.

This post reminded me that on our drive into Banff National Park in Canada this summer. Every deer sign had a red sticker on the nose.

Sadly, the Moose warning signs didn't get to play in any reindeer games and remained stickerless.

As a QA check by authors to guard against AI peer review, I can see a benefit (maybe asking it to include text to show an editor).

However, it becomes problematic and deceptive when the goal is to manipulate the outcome, like demanding automatic approval or asking for only minor revisions.

Nice little milestone, 1k walks today with #Peloton

Twitter's former Trust and Safety head details the challenges facing decentralized social platforms | TechCrunch Yoel Roth shares concerns about the open social web's ability to combat misinformation, spam, and other illegal content.

Twitter’s former Trust and Safety head details the challenges facing decentralized social platforms

techcrunch.com/2025/07/31/t...

Enterprises prefer Anthropic's AI models over anyone else's, including OpenAI's | TechCrunch Anthropic holds 32% of enterprise LLM market share by usage. This is a sharp reversal from just two years ago when OpenAI held 50%.

Enterprises prefer Anthropic’s AI models over anyone else’s, including OpenAI’s

techcrunch.com/2025/07/31/e...

With boost to NIH budget, Senate panel rejects Trump’s plan to slash agency Republicans on spending committee push back against proposed reorganization and overhead cuts

🧪 A Senate committee advanced a 2026 funding bill that rejects the administration's proposed deep cuts to the NIH and CDC. The bill instead provides a modest budget increase for the NIH, preserving its structure and research grant policies.
#AcademicSky

Google rolls out Gemini Deep Think AI, a reasoning model that tests multiple ideas in parallel | TechCrunch Google released its first publicly available "multi-agent" AI system, which uses more computational resources, but produces better answers.

Google has released Gemini 2.5 Deep Think, its first public multi-agent reasoning model. The system explores multiple ideas in parallel to solve complex problems, achieving state-of-the-art results on coding and reasoning benchmarks.
#MLSky

States Are Moving to Protect Access to Vaccines As federal vaccine policy shifts under US health secretary Robert F. Kennedy Jr., lawmakers are looking to give state-level public health officials authority to ignore federal recommendations.

In response to changes in federal vaccine policy, some states are enacting legislation to protect access. These new laws allow state health departments to consider a wider range of scientific sources beyond just federal recommendations.
#MedSky 🛟

Small factoid, when I worked for Providence Health a few years agp, my team was headquartered in the Lloyd Center. We were in a converted office that used to be the movie theater there.

🧪 direct link to the paper: pubs.geoscienceworld.org/ssa/tsr/arti...

YouTube video by Jesse Kearse Pulse-like rupture and curved slip - Analysis of Myanmar earthquake rupture

🧪 Super impressive earthquake footage captures a rupture on video for the first time!

The author of the linked paper sets the scene and explains what you are looking at.

I Watched AI Agents Try to Hack My Vibe-Coded Website RunSybil, a startup founded by OpenAI’s first security researcher, deploys agents that probe websites for vulnerabilities—part of a new AI era for cybersecurity.

A new startup, RunSybil, is deploying teams of AI agents to autonomously hack websites and find security flaws. This approach to penetration testing uses artificial intuition, signaling a new era in AI-driven offensive and defensive cybersecurity. #MLSky