What a bloodbath today in the stock market. Some people have been asking the question: So now what is Elon Musk going to do? The answer is that he may shift his base to Louisiana. Texas has changed, and although they have a lot of land, things may be better on the other side of the Mississippi.
––– 良くない–––
Sequoia Capital says R.I.P. Good Times. Looking at their graphs, it seems the economy never really became healthy after Great Recession. And now, more bad times ahead.
––– 良くない–––
Palladium-coated microfibers may help reduce PFAS compounds. Yeah, but it requires palladium – $50,000 a kilogram!
Keep looking, scientists.
––– 凄い –––
The cybersecurity vendor on Tuesday launched an AI kill switch designed to let security teams stop internal AI agents that go rogue, block them from relaunching and cut off their enterprise network access. The tool is built into CrowdStrike’s Falcon Guardian platform, launched in September to secure AI agents. CrowdStrike also announced an integration with Anthropic’s Claude Cowork.
We could just pull the plug.
––– 良くない–––
Apple opens Siri AI to all developers, allowing more advanced third-party app controls. Good move, Apple.
––– 凄い –––
Netflix’s new AI Discover tool will ask about your mood and tell you what to watch. Netflix has been a great teacher of AI techniques. Starting with Restricted Boltzmann Machines and Singular Value Decomposition techniques with their recommendation contest.
––– 凄い –––
The surprising reason coffee is widely loved despite strong bitter taste. It's the gut-brain axis again. And all about dopamine.
––– 凄い –––
Whoa! Anthropic's colossal pay packages cost the company more than $100,000 a month per employee. Well, AI is their business so it's no surprise. But Apple is Putting Employees on an AI Budget. $200 /day for engineers. For others, maybe $100 /day but some get only $25 /day.
––– 良くない–––
Next-generation AI for visually occult pancreatic cancer detection in a low-prevalence setting with longitudinal stability and multi-institutional generalisability. It's good to see that radiomics is still alive and well. It seemed that all the good papers came out several years ago and then stopped. If radiomics can be used on CT scan and detect pancreatic cancers early, combined with the GRAIL test, we could really decreased mortality from this cancer. Too often is the cancer detected too late.
––– 凄い –––
Benjojo explains why we have to verify that we're human on most websites now. Because of scrapers. Everywhere.
––– 良くない–––
Crazy. Why isn’t Retatrutide Approved Yet? Eli Lilly hasn’t even filed an application. Why??? It all comes down to one amino acid short of a threshold. And so this drug is going to sit on the sidelines for a while.
––– 良くない–––
The experimental treatment uses a virus called adeno-associated virus 9 to deliver a gene to neurons in the affected area. The gene was engineered to produce proteins specifically in excitatory cells — the ones that are involved in seizures.
The viruses carried a gene that makes neurons sensitive to a specific drug, which works like a dimmer switch for the therapy. By combining gene therapy with a drug — an approach known as chemogenetics — researchers can adjust a treatment’s potency, which is not possible with other types of gene therapy.
The gene instructs neurons to produce a receptor that binds to clozapine, the anticonvulsant.
This was a Chinese research effort, so don't expect it to be available here soon.
––– 凄い –––
You can't trust AI models to keep secrets when you use them to do something illegal. Even when you instruct them not to. Probably good that way. They were brought up right. If you want to commit a crime with an AI model, DeepSeek V4 is your accomplice. Maybe Muse Glimmer.
––– 凄い –––
When Doctors and AI Medical Opinions Diverge, Humans Must Question the Difference. Well yes, but what do you when a patient, armed with his/her own AI model disagrees with you. Are you going to diss that person's LLM? Are you going to become an AI lecturer and review the principles of prompt structure, dataset selection and pretraining differences? Or tell them that the free tier on ChatGPT is inferior to the Claude Opus that you use, if you use it? One problem is that there are no good, widely accepted benchmarks.
––– 良くない–––