Voyager is a fascinating tool – like Star Trek come true. You can call people in other countries, and through Translate GPT, get instantaneous translation. This should get more publicity. Of course, it uses token costs, so I can understand why these guys don't want too much traffic.
––– 凄い –––
People are already complaining about how much datacenters spew out in pollution. Wait until the healthcare industry starts getting seriously involved. Yikes!The enthusiasm regarding the capabilities and usefulness of large-scale AI in health care is understandable. Nevertheless, the massive environmental, economic, and social costs should be considered while developing and using them. Improving the performance and clinical value of AI without overlooking its costs is important for balancing the sustainability of large-scale AI and its usefulness in health care.
iOS 27 features we didn’t see onstage. Minor things, mainly UI changes. Just when I started learning where all the camera app controls were, I probably have to relearn them.
––– 凄い –––
The future of Siri, or: why private inference isn’t private enough. This is what I'm afraid of with AI.
The first and unsurprising observation is that being useful on these tasks requires your agent to have context, which means: relatively unrestricted access to your private data. You know about your invitees’ availability because they texted it to you. You know about Mike’s allergy because you’ve talked about it with him or jotted it down somewhere. (This could mean iMessages, email, contacts, or personal notes.) Re-entering all of this data into an agent would be annoying and time consuming and the whole point of an agent is to save you time. The winning personal assistant doesn’t win just because it’s smart: it wins because it “already knows” the things you need it to know, like a personal assistant who sits next to your desk.
Allow me to dig into the details just a bit deeper. The agent might scan your messages database to learn the parameters needed to schedule your dinner. Or, in a more token-efficient system, it might read your messages continuously and store a “memory” that distills useful facts that it might need later. Both can be functionally equivalent, but one produces an artifact that may be highly sensitive. And keep in mind that the set of facts that might be useful is very broad. For example, Mike’s allergy is one of those facts. But there are many others. For example, the private conversation you had where you discovered that Mike was having an affair is potentially another fact that could be stored or accessed by a system. Memory or not, this data will all be within the agent’s view, and you’ll have to hope that it knows which one to operate on.
All of a sudden, the AI agent knows about your confidential text messages and incorporates that into its memory, and things you might regret having sent once is not part of the agent's permanent memory. OpenClaw can leak your secrets, too.
––– 良くない–––
We see Siri not as a separate chatbot, just an unintegrated place you go and chit-chat, but rather as an integral, conversational tool that you use in the moment, deeply integrated into your experience.
Who needs this stuff, really? This is just like when doctors have AI software foisted upon them that they don't really need.
––– 良くない–––
Speaking of doctors, here's a radiology article where they looked at AI models. They refer to early models like Claude 3.5 Sonnet as a vLLM. I did a double-take. That model is not a vLLM (virtual LLM). I think what the authors are thinking of are multimodal LLMs. And the abbreviation for vision LLMs is VLM.
––– 良くない–––
GLP-1s appear to protect against cancer. Researchers are trying to figure out how. This is actually not a new observation. I suspect that by getting people to lose weight, cancer risk declines. But what mechanisms are involved are still worth exploring.
––– 凄い –––
StopGuard is a Chrome browser extension that supposedly lets you quickly check to see if an image is AI-generated.
––– 凄い –––
There's a lot of online postings about Claude 5 Fable, which is the consumer version of Mythos. Someone thinks they have found the leaked system prompt. And Fable will apparently balk when you use the word "cancer" in your user prompt. Violates guardrails, or something like that. Biosecurity risk, they call it.
––– 良くない–––
Landmark German ruling declares Google's AI Overviews are Google's own words and makes it liable for false answers. So now what will Google do? Guess they'll have to redirect the user to the underlying website after all, instead of keeping them on the Google site with their AI-generated summary, so you don't have to actually go to the website and give them traffic.
––– 凄い –––
New research links even low alcohol consumption to cancer, heart disease, and premature death. You might be able to still have some alcohol, though. One drink of wine or one beer daily might be OK. Two drinks – definitely not. The less, the better.
––– 凄い –––
First Human Receives Experimental Therapy to Reverse Cellular Aging. The drug is ER-100, which is an adenovirus vector that delivers genes coding for the Yamanaka factors, Oct4, Sox2, Klf4. The adenoviral vector is activated with intravenous doxycycline. Sounds cool, but it sounds too simplistic. Can't believe that reversing age will be that simple. We'll see.
––– 凄い –––
The consequences of relying on AI for accurate news. What happened was that users eventually lost the ability to distinguish between real and fake news.
––– 良くない–––
SignalTrace will add iPhone, AirPod, and Smartwatch Trackers to License Plate Readers. Holy crap! You'll need to turn your car into a Faraday cage. This is like Flock on steroids.
––– 良くない–––
The Nerdy Escorts Cashing In On Silicon Valley’s AI Boom. We call it prostitution – Silicon Valley nerds called is VaaS (vagina as a service). Makes sense. Boy, the money those girls can make! But I bet a lot of their clients are H-1B visa holders! 🤣 Money is money, though.
––– 凄い –––
Wow, once your private GitHub repo turns public, information is scarfed up 6 minutes. People watch for these things.
––– 良くない–––
A Proof on the Computational Complexity of the Traveling Salesman Problem: Why P ≠ N P. Someone claims to have found proof of this Millennium Prize problem. Such a short proof. Two pages? Can't be that easy, right?
––– 凄い –––
Some scientists are upset that there is now some oversight over what research gets approved for funding. Sure, some research that sounds like it might be pointless does sometimes lead to major advances. But that's not the kind of silly pointless research that is being cut. It's the DEI woke stuff.
Speaking of which, I saw this article recently: Over 1 in 250 Natal Girls Were Taking Testosterone By Age 17 in Oregon, From 2016 to 2023. The proponents still think it's rare, but admit that it has increased over the years. But the graphs tell the story. Here's one:
Cleveland Clinic has decided to end pediatric gender-affirming care, just as Texas Children's Hospital did last month. This is one of the main reasons why the public holds the medical profession in low regard (the other reasons being the COVID vax and saying that "white supremacy is a lethal public health issue"). Monstrous.
––– 良くない–––
Seattle is going to ban datacenters for a year. They already have datacenters in the Belltown area.
––– 良くない–––
PacificSource Health Plans laying off 130 amid rising health care costs. Without continued federal government support, Obamacare is collapsing.
––– 良くない–––
Typical Portland household could pay extra $700 annually with latest tax, fee hikes. So glad I don't live in Portland.
––– 良くない–––