Shrew
@shrewshrew.bsky.social
4K followers 800 following 2.7K posts
Neuroscientist/mouse psychologist and mad about it. She
Posts Media Videos Starter Packs
Pinned
shrewshrew.bsky.social
Today I am wondering about the insights we could have about the brain if we didn't insist that it recapitulates western power structures

For example, to what degree are our hypotheses about prefrontal cortex limited because we call its processes "executive functions" instead of like, "mothering"
shrewshrew.bsky.social
As the prophet once said
It is Wednesday my dudes
AAAAAAAAAAHHHHHHH
shrewshrew.bsky.social
The Naturalistic Truthaccy
Reposted by Shrew
writesantonio.bsky.social
Why is the economy still doing relatively well? We invested money in AI infrastructure which I’m sure has a few loose nuts and bolts on its legs 👀
justinhendrix.bsky.social
"The hundreds of billions of dollars companies are investing in AI now account for an astonishing 40 per cent share of US GDP growth this year... In a way, then, America has become one big bet on AI."
America is now one big bet on AI
It’s seen as the magic fix for every threat to the US economy
www.ft.com
shrewshrew.bsky.social
Why is it so humid. I thought I left humidity behind when I moved to the mainland. why is it obsessed with me
Reposted by Shrew
jam2go.com
I’ve been training my cat to play ambient music.

Last night he hit me with this hauntingly beautiful chord change after 10 minutes of A7sus4
Reposted by Shrew
clowndro.bsky.social
[letter to vitamin company]

So your ad said that this supplement "helps with memory loss" but I'm still remembering stuff
Reposted by Shrew
mindflakes.bsky.social
My many enemies will be furious to know that I found a luxurious new hat at the bus station
shrewshrew.bsky.social
I do gotta hand it to them that they at least decided to ask the question "can it understand information" rather than interacting with AI billionaire style, eg "it instantly understands all human knowledge including complicated science almost no one understands and would be my friend if I had those"
Reposted by Shrew
kurtfraser.bsky.social
This is actually who is behind ChatGPT. Just thousands and thousands of layers of rodents on their phones
shrewshrew.bsky.social
Why is this the first I am hearing about rodents having thumbnails??

Well anyways if they have thumbs we can give them tiny phones and tiny social medias and maybe tiny sports betting apps. No better mousetrap has ever been devised

@smithsonianmag.bsky.social www.smithsonianmag.com/smart-news/r...
Rodents Conquered the World With the Help of Their Thumbnails, Study Suggests
The trait might have given rodents greater manual dexterity, allowing them to access new foods, such as nuts
www.smithsonianmag.com
shrewshrew.bsky.social
Oh that is a legitimate great tip ty
Reposted by Shrew
ampolter.bsky.social
Smh. Taking good middle class jobs from comically antisocial confident lieHUMANS
shrewshrew.bsky.social
Ok how do I turn this into an op ed
shrewshrew.bsky.social
I teach about how casinos and gambling apps drive compulsive use via providing the right rate of "close enough" payoffs to destroy peoples finances and lives for profit

and it honestly feels like such childs play compared to AI companies running this playbook to destroy fact itself
shrewshrew.bsky.social
Univeraally the wisest and most chill kind of human
shrewshrew.bsky.social
shrewshrew.bsky.social
It is *pleasurable* to be certain about what we should do. Our brain enters a different state. We like it so much we spend time and money to keep doing that activity.

Uncertainty, thinking, failing? they can't compete.

A real problem is: at their core, AIs are slot machines where you win certainty
metr.org
METR @metr.org · Jul 10
When AI is allowed, developers spend less time actively coding and searching for information, and instead spend time prompting AI, waiting on/reviewing AI outputs, and idle. We find no single reason for the slowdown—it’s driven by a combination of factors.
Reposted by Shrew
shrewshrew.bsky.social
I teach about how casinos and gambling apps drive compulsive use via providing the right rate of "close enough" payoffs to destroy peoples finances and lives for profit

and it honestly feels like such childs play compared to AI companies running this playbook to destroy fact itself
shrewshrew.bsky.social
I teach about how casinos and gambling apps drive compulsive use via providing the right rate of "close enough" payoffs to destroy peoples finances and lives for profit

and it honestly feels like such childs play compared to AI companies running this playbook to destroy fact itself
shrewshrew.bsky.social
We will repetitively, compulsively, make choices that make our lives simpler and nothing is simpler than being right 😎

Our brave silicon valley overlords have simply found a way to offer the feeling of being right without worrying about any pesky "actually being right"
shrewshrew.bsky.social
Now that it is clear that AI benchmarks and training algorithms penalize the AI for not being confident, it's essential to understand the psychology of user decision making that explains why these companies would do something so comically antisocial as build planet-destroying confident liebots
shrewshrew.bsky.social
It is *pleasurable* to be certain about what we should do. Our brain enters a different state. We like it so much we spend time and money to keep doing that activity.

Uncertainty, thinking, failing? they can't compete.

A real problem is: at their core, AIs are slot machines where you win certainty
metr.org
METR @metr.org · Jul 10
When AI is allowed, developers spend less time actively coding and searching for information, and instead spend time prompting AI, waiting on/reviewing AI outputs, and idle. We find no single reason for the slowdown—it’s driven by a combination of factors.
shrewshrew.bsky.social
Don't make me tap the sign

Large language models are slot machines that pay out in confidence, not information

If the information is bad (it often is) no one cares

But if the confidence was bad no one would use it 🤔

Luckily for the compulsive use companies want to promote, it's always confident
swilua.bsky.social
you don’t understand bro, if we didn’t lie, no one would buy our product 😩
Consider the implications if ChatGPT started saying "I don't know" to even 30% of queries
- a conservative estimate based on the paper's analysis of factual uncertainty in training data. Users accustomed to receiving confident answers to virtually any question would likely abandon such systems rapidly.
Reposted by Shrew
swilua.bsky.social
you don’t understand bro, if we didn’t lie, no one would buy our product 😩
Consider the implications if ChatGPT started saying "I don't know" to even 30% of queries
- a conservative estimate based on the paper's analysis of factual uncertainty in training data. Users accustomed to receiving confident answers to virtually any question would likely abandon such systems rapidly.
Reposted by Shrew
jessicacalarco.com
Parents *shouldn't have to* lead the way in helping kids navigate AI risks and opportunities. Parents should be able to trust that developers and regulators are working to ensure that AI is safe and ethical, and that educators are trained to help kids navigate these tools responsibly. 1/
Reposted by Shrew
internethippo.bsky.social
Guy who sells doo doo (squinting wisely and doing visionary voice): We believe within 10 years the world economy will be almost completely powered by doo doo
Reposted by Shrew
arielron.bsky.social
Graphic summary of the "tyranny of the wagon equation" for a Civil War army of 100,000 men + 20,000 cavalry & artillery horses + X wagon mules as a function of days' march from a base. The equine supply factor is total daily forage ÷ food (for humans). By day 4 or 5 it's out of control.