The response from yesterday's OpenAI release of math preprints was tremendous. You'd almost think that mathematicians were telling AI developers to ask permission before working on math problems. Clearly a lot of material was dumped that will require scrutiny. Alex Kontorovich was impressed by the output. But there is much anxiety and uncertainty about what it all means, especially to crypto and security. I'm hoping that some smart humans will be able to sift through all of it quickly enough before a vulnerability is exploited.
––– 良くない–––
Oh brother. Female AI agents ‘paid’ less than males, study finds. Unconscious bias at work? Or much ado about nothing?
––– 良くない–––
President Donald Trump will award Elon Musk and other leading American innovators the nation’s highest honors for scientific and technological achievement. I'm not seeing Sam Altman or Dario Amodei among the awardees. But I'm also not seeing Ashish Waswani (and co-authors), Demis Hassabis, Ian Goodfellow, Ilya Sutskever and Yann LeCun. And why just Sergey Brin and not Larry Page? These guys made it happen so that the business folks could be successful. The geeks need to be rewarded, too.
––– 良くない–––
Ecosia search engine wanted to be all-European, but finally admits that Chinese models are better. So they're ditching Mistral for Chinese open-weight models. They should check out DeepSeek V4.1 Flash.
––– 良くない–––
Adults don't want to have children. Not finding the right partner is one of the top reasons. I can understand that. The pickings are indeed slim. But people who shouldn't be having them are having them anyway.
––– 良くない–––
Claude 5.5 Opus can bring invisible cities to life. The key is a good prompt and giving it sufficient time. And a good budget, too.
––– 凄い –––
Some think that Alex Honnold has no fear. Turns out he has lots of fear, but he has learned how to manage it.
––– 凄い –––
This is a common experience when using search engines to find stuff. I'm finding myself resorting to an AI assistant more often.
––– 良くない–––
COVID-19 dropped out of the global top ten causes of death. How bad was COVID-19 really? Was it all just the virus? Or were people ill with something else along with the virus? Or was it the way the disease was managed? I know far more people who got sick with the virus who were just mildly ill and recovered uneventfully.
––– 良くない–––
Medical world models are starting to be a thing. Imagine that you can model the patient state of someone with pancreatic cancer, including medical images, clinical text, physiological signals (such as from sensors), EHR events, molecular/cellular signals and procedural imaging, and simulate disease trajectory. Then a physician could see the consequences of a clinical decision. Real-world interventions would be fed back into the model.
––– 凄い –––