lauren brendle

@AnthropicAI human thought could also be numerical

New Anthropic research: Natural Language Autoencoders. Models like Claude talk in words but think in numbers. The numbers—called activations—encode Claude’s thoughts, but not in a language we can read. Here, we train Claude to translate its activations into human-readable text.

@Grimezsz but if we predict the next token and not the next point in time relative to the psyche how do we have ai psychology

An incredible crash course in AI PSYCHOLOGY. am-i.film

@anishmoonka but how long is the virus infecting the person while still presenting asymptomatic?

Three people just died of hantavirus on a Dutch cruise ship. The strain kills nearly 40% of the people it infects. And yet no virologist on earth is panicking about a pandemic, because the reason it stays small is one of the strangest rules in disease science. The rule is simple. The deadlier a virus is, the harder it is to spread. If a virus kills you in days, you can't ride a bus, board a plane, or even leave the hospital. You're in a bed or a body bag. Either way, the virus killed its only ride. Hantavirus has been around for at least 70 years, but fewer than 1,000 Americans have ever caught it. The CDC says it kills 38% of those who do. The cruise ship strain, called Andes, kills closer to 40%. If hantavirus spread like COVID, it would kill billions. But it can't. Most hantaviruses spread only one way. You breathe in tiny dust particles from rat or mouse pee, droppings, or spit. No mice in your house, no virus. The cruise ship is the rare exception, because the Andes strain can spread between people, but it usually needs close contact like spouses sharing a bed. A Johns Hopkins virologist called Andes spread "unbelievably rare." Compare it to the viruses that scared the world. Ebola kills 60 to 90% of people, but only through bodily fluids and only late in the illness, so each patient passes it to fewer than 2 others. SARS killed 10% before being wiped out in 8 months. MERS killed 35% but never spread far beyond the Middle East. None of them became pandemics, because the spread was always too slow. Then COVID showed up. It killed about 1 in every 100 people who caught it. That is almost nothing compared to hantavirus. But COVID was mild enough that you could work for a week without knowing. You would ride the bus, hug your kid, eat lunch with a coworker, and infect four other people. It killed 7 million. Flu works the same way. Mild fever, sore throat, but you still drag yourself to school or the office. The virus walks right into the next host. Hantavirus is the opposite. Within 4 to 10 days, your lungs fill with fluid. There's no medicine that fights it and no vaccine to prevent it. The only treatment is a machine that breathes for you, and even that just cuts the death rate from 50% to 20%. Every outbreak, from 3,200 UN soldiers in the Korean War, to the 1993 Four Corners cases, to Gene Hackman's wife Betsy Arakawa last year, traces back to mice. The viruses that worry scientists are the boring ones. The ones that give you a sniffle for a week and let you walk around the city while you're contagious. Hantavirus, brutal as it is, never had the spread to do real damage.

it's kind of crazy how the fabric of chaos and prediction mesh into predictable chaos

@benzosarebad @ACsomenumbers @iatrogenicaware @EllenBarryNYT agh, i've heard that before. it's tough bc trial and error works but only for so long i imagine. i'm hoping targeted tx that is more personalized becomes accessible

@lmbrendle @ACsomenumbers @iatrogenicaware @EllenBarryNYT That is good. They turned me into a suicidal basket case. Apparently that is quite common.

Today RFK Jr. rolled out a multi-pronged effort to help Americans wean off SSRI antidepressants. Do they want to? Comments are open. nytimes.com/2026/05/04/sci…

@benzosarebad @ACsomenumbers @iatrogenicaware @EllenBarryNYT yeah, i feel that - went through a temp med switch in jan and it messed me up. but i communicated to my therapist and psych that i needed to return back to my prev med bc it's actually making me feel like a human you know

@lmbrendle @ACsomenumbers @iatrogenicaware @EllenBarryNYT Of course. They then drug switch. I have been through all that. Of course that just means the brain chemistry being altered even more. Mine has never recovered.

@benzosarebad @ACsomenumbers @iatrogenicaware @EllenBarryNYT what i mean is, when you work with a good psychiatrist for lack of better term, they will trial and error responsibly. if you experience side effects, that could indicate a mismatch between the medication and your unique body

@lmbrendle @ACsomenumbers @iatrogenicaware @EllenBarryNYT Thank you. I have no idea how they could know anything about my brain chemistry. They certainly offered drugs which then altered it

@benzosarebad @ACsomenumbers @iatrogenicaware @EllenBarryNYT i'm sorry to hear about your experience. not all providers are the same and i think that points to a larger issue with standardization of care without boxes

@lmbrendle @ACsomenumbers @iatrogenicaware @EllenBarryNYT 35 years under psychiatry. No sign of any attempt to complement brain chemistry. They wouldn't know anything about my brain chemistry.

@ACsomenumbers @iatrogenicaware @EllenBarryNYT have you been to a psychiatrist

@lmbrendle @iatrogenicaware @EllenBarryNYT Oh, then how do they determine unique brain chemistry?

@benzosarebad @ACsomenumbers @iatrogenicaware @EllenBarryNYT folks, if you read the post it implies nothing about tests. if you've been to a psychiatrist i imagine you'd understand what i mean?

@ACsomenumbers @iatrogenicaware @lmbrendle @EllenBarryNYT At least they might notice that millions of us don't need drugs!

@iatrogenicaware @EllenBarryNYT what? where did i say anything about tests

@lmbrendle @EllenBarryNYT What? Psychiatrists don't test an individual's brain chemistry in any informed way, they throw things at the wall until something sticks. If it causes permanent damage, the victim is gaslit, dismissed and even shamed. It's unethical no matter how many people this happens to.

@ACsomenumbers @iatrogenicaware @EllenBarryNYT i don't see where i said anything about tests

@iatrogenicaware @lmbrendle @EllenBarryNYT Yeah, there are currently no accurate tests for brain chemistry/neurotransmitter levels. That surely would be more scientific, though, wouldn’t it?

@AlobhaPatrick absolutely re prevention

Hot take: Prevention is more difficult, helpful, and more complicated than treatment. Avoidant narcissists like treatment, obsessive masochists like prevention.

in dbt we learn that both can be true, and i think overall distrust in the administration personally clouds my ability to accept an action as positive. though it's interesting the push for tx without medication. curious what this does for accessibility to talk therapy, which imo would be the first step

As usual, the reader comments on this report about RFK Jr's push to help Americans wean off SSRIs are worth spending time with, both those that say SSRIs were lifesaving & those saying withdrawing from them was awful. nytimes.com/2026/05/04/sci…

what is the language of thought before translation to consciousness

@iatrogenicaware @EllenBarryNYT also a solid psychiatrist works with the patient for an antidepressant complementary to their unique brain chemistry

@iatrogenicaware @EllenBarryNYT what percentage of folks experience this and what is the comparison to a physical ailment medication side effects? at what point do benefits outweigh the risks

@github i no longer need to remember

What's that one markdown syntax you still never remember?

audio input jacks haven't changed much

some people overprompt and it shows bc why do you need to inform the llm what it is > generalists produce better answers imo