On the difference between real emergencies and pretend emergencies:
One day I was sitting on the back porch while my 2 1/2 yo upstairs neighbor was playing in his sandbox on the other side of the driveway. Suddenly he got up and ran across the back yard, where his mom was weeding her garden, shouting "Mom! Mom! There's an emergency going on!" Knowing her son well, she asked "Is it a real emergency or a pretend emergency?" He replied "It's a REAL pretend emergency! The dump truck is buried in the sand." Mom explained "Ben, that's a PRETEND emergency."
Having less understanding than an imaginative and articulate two year old, the Trump administration is incapable of distinguishing real from pretend emergencies, at least when it is inconvenient to do so. We do live in the stupidest possible reality.
Current pseudo AI tech (really, very advanced pattern matching) is not like Asimov's robotics. There is no hierarchy of laws. Any rule can be violated, including Asimov's First Law.
There are other types of AI, including rule-based AI, that cause the AI to apply rules (either programmed in or discovered by experience) to a situation to discover the best solution to a problem, or at least to evaluate various solutions to a problem so they can be compared. The current crop of LLM-based AI's do NOT do this.
Instead, they examine an extremely large database of text (or of images for creating or analyzing pictorial data) and predicate what is most likely to come next. These databases include facts, lies, mistakes, misconceptions, reality, fiction, and anything else the web crawler finds. The AI's have no way of distinguishing truth from falsity, fact from fiction, physics from magic, or any other sort of reality-checking. Some people claim this is unnecessary, that LLM's are good enough. I think this is suicidal.
My dumb state of Florida is invoking emergency rules to keep gay people from getting HIV medication. Yep. They did that.
They're taking away funding for Biktarvy, for which you probably have seen ads on TV. A popular medication that does what it says, and they're it away, just because they don't like gay people.
The emergency rules can be renewed every 90 days.
Know what is the correct amount of time to put up with this stupid bullshit?
0 days. Precisely 0.
Surgeon General Joseph Ladapo said the cut was made because of a drop in federal funding, but AIDS patient advocates say that is not the case.
OMFG. Are you goddamned kidding me?!?!?!? Yes, I see those ads on the TV and am glad they have those medications to keep folks from getting HIV and AIDS.
Thank you, Birb, for letting us know, and for volunteering!
In our basement rec room. CBS, NBC & ABC. No PBS cause UHF with terrible signal, and not in our broadcast area. Channel surfing was getting up off the couch and manually turning, or Don't Turn That Dial on the b&w 12 feet away. It was easy.
When I worked for Raytheon this guy who was given the title of Engineer and every fucking day would come in and assail everyone with Rush's latest wisdom, would tell us if Clinton were elected and taxes went up we'd lose our jobs because no one would buy planes because taxes. Of course I tried to explain the foolishness in his blanket statement and he stuck to his guns anyway. I wish I'd been around eight years later to laugh in that ogre's face.
For a long time it has been a concern that eventually AI could harm humans, but safeguards apparently were being built into AI systems to prevent AI from having this power. AI will not harm humans, humans are harming humans with the AI system they created to harm humans, among other things that make life worse, not better.
This says a lot: Anthropic’s core principle was to “…pause training more powerful models if their capabilities outstripped the company’s ability to control them and ensure their safety.” Not OUR safety, not humanity or America or even the Anthropic CEO’s safety. The MODEL’S safety. Wouldn’t want it to develop some sort of electronic neurosis after nuking the planet…
No, no LLM is good for running weapons systems. It's just not a good idea. Honestly, I'd be quite happy to see the whole field abandoned as a dead end.
Apparently, the WARFIGHTERS were using the AIs for data analysis. AI in weapon systems won't ever make it out of testing. Even Anduril, the AI weapon company, isn't using AI for targeting ... just navigation.
I'm working on the computer administered version of a test and I keep having to tell the authors that the "computer" won't score any of the test. The examiner scores the items and records the scores in the program. The authors are really having a hard time understanding why we cannot just use AI and I have to be like, "Because AI isn't trained for this."
Not how I read that. As company, they have been one of two "AI" model training companies that test their models for safety. The other is Mistral from France.
I guess you could read it as “…ensure their safety as a tool to be used by humanity.” But if you have to parse it cautiously and make assumptions to understand it, they could have phrased it better.
Damnnn! This gotta be a one hell of a read.
On the difference between real emergencies and pretend emergencies:
One day I was sitting on the back porch while my 2 1/2 yo upstairs neighbor was playing in his sandbox on the other side of the driveway. Suddenly he got up and ran across the back yard, where his mom was weeding her garden, shouting "Mom! Mom! There's an emergency going on!" Knowing her son well, she asked "Is it a real emergency or a pretend emergency?" He replied "It's a REAL pretend emergency! The dump truck is buried in the sand." Mom explained "Ben, that's a PRETEND emergency."
Having less understanding than an imaginative and articulate two year old, the Trump administration is incapable of distinguishing real from pretend emergencies, at least when it is inconvenient to do so. We do live in the stupidest possible reality.
"The Trump Pentagon, run by inveterate liars, says no, no, we have it all wrong."
Additionally: invertebrate liars.
"Welcome to dystopia[,]" indeed!
Sea slug libelz
Can't they imbed AI that adopts the minimum obligation that any illegal orders cannot be carried out? Just me and Mark Kelly etc asking.
Current pseudo AI tech (really, very advanced pattern matching) is not like Asimov's robotics. There is no hierarchy of laws. Any rule can be violated, including Asimov's First Law.
There are other types of AI, including rule-based AI, that cause the AI to apply rules (either programmed in or discovered by experience) to a situation to discover the best solution to a problem, or at least to evaluate various solutions to a problem so they can be compared. The current crop of LLM-based AI's do NOT do this.
Instead, they examine an extremely large database of text (or of images for creating or analyzing pictorial data) and predicate what is most likely to come next. These databases include facts, lies, mistakes, misconceptions, reality, fiction, and anything else the web crawler finds. The AI's have no way of distinguishing truth from falsity, fact from fiction, physics from magic, or any other sort of reality-checking. Some people claim this is unnecessary, that LLM's are good enough. I think this is suicidal.
The more ai slop that infests the intertubes, the more ai trained on the Internet becomes unusable
Why 5:01? Why not plain five o'clock?
I'm absolutely sure it's been noted earlier, but isn't this the prequel to "The Terminator"?
OT:
Hi, good morning, everyone.
Get ready to throw things at the wall.
My dumb state of Florida is invoking emergency rules to keep gay people from getting HIV medication. Yep. They did that.
They're taking away funding for Biktarvy, for which you probably have seen ads on TV. A popular medication that does what it says, and they're it away, just because they don't like gay people.
The emergency rules can be renewed every 90 days.
Know what is the correct amount of time to put up with this stupid bullshit?
0 days. Precisely 0.
Surgeon General Joseph Ladapo said the cut was made because of a drop in federal funding, but AIDS patient advocates say that is not the case.
https://www.cbsnews.com/miami/news/florida-department-of-health-cut-hiv-aids-drug-program/
If you want to make a complaint, here is the link to the form to do that. Please read the article, absorb the full information, and send the message.
https://www.floridahealth.gov/about-us/contact-us/
I have a local congressman to get pissed off at.
Birb is going to volunteering soon, but I will take care of this toutes de suite (immediately).
OMFG. Are you goddamned kidding me?!?!?!? Yes, I see those ads on the TV and am glad they have those medications to keep folks from getting HIV and AIDS.
Thank you, Birb, for letting us know, and for volunteering!
Thank you.
I didn't get good sleep, so I'll need a nap after I come home, but it was good to find out about that and let everyone know.
Fuck's sake, they don't even do consent for the killer robots
I bet Palantir is behind this, they want to seize Claude.
Yes
Vance has always been Thiel's puppet
And Theil has always been a strange twist
Yes. Palantir uses Anthropic's tools in their products a lot.
Hastening the demise of legacy media.
𝘗𝘦𝘳 𝘵𝘩𝘦 𝘵𝘦𝘳𝘮𝘴 𝘰𝘧 𝘵𝘩𝘦 𝘰𝘳𝘪𝘨𝘪𝘯𝘢𝘭 𝘥𝘦𝘢𝘭, 𝘞𝘢𝘳𝘯𝘦𝘳 𝘉𝘳𝘰𝘴. 𝘋𝘪𝘴𝘤𝘰𝘷𝘦𝘳𝘺 𝘸𝘪𝘭𝘭 𝘩𝘢𝘷𝘦 𝘵𝘰 𝘱𝘢𝘺 𝘢 $2.8 𝘣𝘪𝘭𝘭𝘪𝘰𝘯 𝘵𝘦𝘳𝘮𝘪𝘯𝘢𝘵𝘪𝘰𝘯 𝘧𝘦𝘦 𝘵𝘰 𝘕𝘦𝘵𝘧𝘭𝘪𝘹 𝘵𝘰 𝘦𝘯𝘥 𝘵𝘩𝘦 𝘦𝘹𝘪𝘴𝘵𝘪𝘯𝘨 𝘢𝘨𝘳𝘦𝘦𝘮𝘦𝘯𝘵. 𝘗𝘢𝘳𝘢𝘮𝘰𝘶𝘯𝘵’𝘴 𝘳𝘦𝘯𝘦𝘸𝘦𝘥 𝘰𝘧𝘧𝘦𝘳 — 𝘣𝘢𝘤𝘬𝘦𝘥 𝘣𝘺 𝘵𝘩𝘦 𝘸𝘰𝘳𝘭𝘥’𝘴 𝘴𝘪𝘹𝘵𝘩-𝘳𝘪𝘤𝘩𝘦𝘴𝘵 𝘱𝘦𝘳𝘴𝘰𝘯, 𝘖𝘳𝘢𝘤𝘭𝘦’𝘴 𝘦𝘹𝘦𝘤𝘶𝘵𝘪𝘷𝘦 𝘤𝘩𝘢𝘪𝘳, 𝘢𝘯𝘥 𝘋𝘢𝘷𝘪𝘥 𝘌𝘭𝘭𝘪𝘴𝘰𝘯’𝘴 𝘧𝘢𝘵𝘩𝘦𝘳, 𝘓𝘢𝘳𝘳𝘺 𝘌𝘭𝘭𝘪𝘴𝘰𝘯 — 𝘪𝘯𝘤𝘭𝘶𝘥𝘦𝘴 𝘱𝘢𝘺𝘪𝘯𝘨 𝘵𝘩𝘢𝘵 𝘣𝘳𝘦𝘢𝘬𝘶𝘱 𝘧𝘦𝘦.
𝘛𝘩𝘦 𝘯𝘦𝘸 𝘥𝘦𝘢𝘭 𝘸𝘪𝘭𝘭 𝘴𝘦𝘦 𝘗𝘢𝘳𝘢𝘮𝘰𝘶𝘯𝘵, 𝘸𝘩𝘪𝘤𝘩 𝘸𝘢𝘴 𝘣𝘰𝘶𝘨𝘩𝘵 𝘫𝘶𝘴𝘵 𝘭𝘢𝘴𝘵 𝘺𝘦𝘢𝘳 𝘣𝘺 𝘌𝘭𝘭𝘪𝘴𝘰𝘯’𝘴 𝘚𝘬𝘺𝘥𝘢𝘯𝘤𝘦 𝘔𝘦𝘥𝘪𝘢 𝘸𝘪𝘵𝘩 𝘩𝘦𝘢𝘷𝘺 𝘧𝘪𝘯𝘢𝘯𝘤𝘪𝘢𝘭 𝘣𝘢𝘤𝘬𝘪𝘯𝘨 𝘧𝘳𝘰𝘮 𝘩𝘪𝘴 𝘧𝘢𝘵𝘩𝘦𝘳, 𝘢𝘤𝘲𝘶𝘪𝘳𝘪𝘯𝘨 𝘵𝘩𝘦 𝘦𝘯𝘵𝘪𝘳𝘦𝘵𝘺 𝘰𝘧 𝘞𝘢𝘳𝘯𝘦𝘳 𝘉𝘳𝘰𝘴. 𝘋𝘪𝘴𝘤𝘰𝘷𝘦𝘳𝘺, 𝘪𝘯𝘤𝘭𝘶𝘥𝘪𝘯𝘨 𝘪𝘵𝘴 𝘴𝘵𝘶𝘥𝘪𝘰𝘴, 𝘏𝘉𝘖, 𝘪𝘵𝘴 𝘴𝘵𝘳𝘦𝘢𝘮𝘪𝘯𝘨 𝘴𝘦𝘳𝘷𝘪𝘤𝘦, 𝘪𝘵𝘴 𝘨𝘢𝘮𝘦𝘴 𝘢𝘯𝘥 𝘦𝘯𝘵𝘦𝘳𝘵𝘢𝘪𝘯𝘮𝘦𝘯𝘵 𝘥𝘪𝘷𝘪𝘴𝘪𝘰𝘯𝘴, 𝘢𝘯𝘥 𝘭𝘪𝘯𝘦𝘢𝘳 𝘵𝘦𝘭𝘦𝘷𝘪𝘴𝘪𝘰𝘯 𝘯𝘦𝘵𝘸𝘰𝘳𝘬𝘴 𝘭𝘪𝘬𝘦 𝘊𝘕𝘕, 𝘛𝘉𝘚, 𝘛𝘕𝘛, 𝘋𝘪𝘴𝘤𝘰𝘷𝘦𝘳𝘺, 𝘢𝘯𝘥 𝘏𝘎𝘛𝘝.
𝗡𝗲𝘁𝗳𝗹𝗶𝘅 𝗯𝗮𝗰𝗸𝘀 𝗼𝘂𝘁 𝗼𝗳 𝗯𝗶𝗱 𝗳𝗼𝗿 𝗪𝗮𝗿𝗻𝗲𝗿 𝗕𝗿𝗼𝘀. 𝗗𝗶𝘀𝗰𝗼𝘃𝗲𝗿𝘆, 𝗴𝗶𝘃𝗶𝗻𝗴 𝘀𝘁𝘂𝗱𝗶𝗼𝘀, 𝗛𝗕𝗢, 𝗮𝗻𝗱 𝗖𝗡𝗡 𝘁𝗼 𝗘𝗹𝗹𝗶𝘀𝗼𝗻-𝗼𝘄𝗻𝗲𝗱 𝗣𝗮𝗿𝗮𝗺𝗼𝘂𝗻𝘁
https://techcrunch.com/2026/02/26/netflix-warner-bros-discovery-paramount-wbd-bid-studios-hbo-cnn-ellison/
In our basement rec room. CBS, NBC & ABC. No PBS cause UHF with terrible signal, and not in our broadcast area. Channel surfing was getting up off the couch and manually turning, or Don't Turn That Dial on the b&w 12 feet away. It was easy.
goddamn it
Fascist Enshitification.
Ellison is the new Murdoch
The number of people you could house for that much money.
The rich are just paying each other. They don't need us anymore
When that crashes we all fall down.
There needs to be limits on how much money a corporation can horde.
"If Mamdani is elected, rich people will flee."
o rly?
https://xcancel.com/jesskcoleman/status/2026755257280119043
This just in: rich people still live in the city of Noo Yawk
I so want him to succeed. Irrationally so.
Trust me, that’s rational.
When I worked for Raytheon this guy who was given the title of Engineer and every fucking day would come in and assail everyone with Rush's latest wisdom, would tell us if Clinton were elected and taxes went up we'd lose our jobs because no one would buy planes because taxes. Of course I tried to explain the foolishness in his blanket statement and he stuck to his guns anyway. I wish I'd been around eight years later to laugh in that ogre's face.
They always threaten to take their ball and go home. I'm glad the voters realized they can't take the demand with them.
Currently, Mayor Mamdani has the highest approval rating of ANY politician in the United States.
They never do.
For a long time it has been a concern that eventually AI could harm humans, but safeguards apparently were being built into AI systems to prevent AI from having this power. AI will not harm humans, humans are harming humans with the AI system they created to harm humans, among other things that make life worse, not better.
"... but safeguards apparently were being built into AI systems to prevent AI from having this power."
Not true, never was true, no longer even an aspiration.
Morning Kitten
https://bsky.app/profile/democratcats.bsky.social/post/3mfoq2g3x222j
Randy Rainbow "The Fate of Melania":
https://www.youtube.com/watch?v=-4g-jFN4yAs
This says a lot: Anthropic’s core principle was to “…pause training more powerful models if their capabilities outstripped the company’s ability to control them and ensure their safety.” Not OUR safety, not humanity or America or even the Anthropic CEO’s safety. The MODEL’S safety. Wouldn’t want it to develop some sort of electronic neurosis after nuking the planet…
I don't care why they do it. If their reasoning keeps them from giving Hegseth what he wants, that's what matters.
Oh, yes. Anthropic's Claude toolset and LLMs are very good for the kinds of things Drunk Petey needs. I hope he cries.
AA would label him a ‘dry drunk’. Look it up, see if you agree.
No, no LLM is good for running weapons systems. It's just not a good idea. Honestly, I'd be quite happy to see the whole field abandoned as a dead end.
Apparently, the WARFIGHTERS were using the AIs for data analysis. AI in weapon systems won't ever make it out of testing. Even Anduril, the AI weapon company, isn't using AI for targeting ... just navigation.
So much of this AI is overblown hype.
I'm working on the computer administered version of a test and I keep having to tell the authors that the "computer" won't score any of the test. The examiner scores the items and records the scores in the program. The authors are really having a hard time understanding why we cannot just use AI and I have to be like, "Because AI isn't trained for this."
Not how I read that. As company, they have been one of two "AI" model training companies that test their models for safety. The other is Mistral from France.
I guess you could read it as “…ensure their safety as a tool to be used by humanity.” But if you have to parse it cautiously and make assumptions to understand it, they could have phrased it better.
Agreed. Technical people (Amodei is an engineer) aren't great at making the message clear.
They probably had their LLM do it.
Maybe they had AI write it ;)
𝟭𝟱 𝘀𝘁𝗮𝘁𝗲 𝗮𝘁𝘁𝗼𝗿𝗻𝗲𝘆𝘀 𝗴𝗲𝗻𝗲𝗿𝗮𝗹 𝘀𝘂𝗲 𝗥𝗙𝗞 𝗝𝗿. 𝗼𝘃𝗲𝗿 “𝗮𝗻𝘁𝗶-𝘀𝗰𝗶𝗲𝗻𝗰𝗲” 𝘃𝗮𝗰𝗰𝗶𝗻𝗲 𝗽𝗼𝗹𝗶𝗰𝘆
Trump administration’s reduced vaccine schedule “throws science out the window.”
https://arstechnica.com/health/2026/02/15-state-attorneys-general-sue-rfk-jr-over-anti-science-vaccine-policy/
How do they write that many words on the subject without listing the states involved?
Only 15? Kind of pathetic.