Lightnews — Scholar-powered news

Carl T. Bergstrom

@carlbergstrom.com

140K followers 2.5K following 9.8K posts

Professor, UW Biology / Santa Fe Institute I study how information flows in biology, science, and society. Book: *Calling Bullshit*, http://tinyurl.com/fdcuvd7b LLM course: https://thebullshitmachines.com Corvids: https://tinyurl.com/mr2n5ymk he/him .. more

Videos

Carl Theodore Bergstrom is a theoretical and evolutionary biologist and a professor at the University of Washington in Seattle, Washington, United States. Bergstrom is a critic of low-quality or misleading scientific research. He is the co-author of a book on misinformation called Calling Bullshit: The Art of Skepticism in a Data-Driven World and teaches a class by the same name at University of Washington. .. more

Biology 21%

Mathematics 20%

Posts Media Videos Starter Packs

Pinned

Carl T. Bergstrom @carlbergstrom.com · Feb 4

Modern-Day Oracles or Bullshit Machines?

Jevin West (@jevinwest.bsky.social) and I have spent the last eight months developing the course on large language models (LLMs) that we think every college freshman needs to take.

thebullshitmachines.com

INTRODUCTION

thebullshitmachines.com

160 950 2.6K

Carl T. Bergstrom @carlbergstrom.com · 59m

I should’ve known you were one of the doesn’t-know-what-ad-hominem-means brigade.

Later, fucknuts.

Carl T. Bergstrom @carlbergstrom.com · 1h

Lol. Given that you don’t even see the problem, it might be a really good time to STFU.

1 9

Carl T. Bergstrom @carlbergstrom.com · 2h

Separate diagnosis with a much higher rate of incidence.

Carl T. Bergstrom @carlbergstrom.com · 2h

Sam Altman: "Now that we have been able to mitigate the serious mental health issues and have new tools, we are going to be able to safely relax the restrictions in most cases."

I'd like to know what they've done to mitigate AI-induced psychosis, which I think is far more common that reported here.

Can AI chatbots trigger psychosis? What the science says

Chatbots can reinforce delusional beliefs, and, in rare cases, users have experienced psychotic episodes.

www.nature.com

11 29 160

Carl T. Bergstrom @carlbergstrom.com · 5h

I got nothin'.

The Verge @theverge.com · 9h

Sam Altman says ChatGPT will soon sext with verified adults

ChatGPT is about to get flirty.

buff.ly

14 15 110

Carl T. Bergstrom @carlbergstrom.com · 5h

I have no idea what you’re talking about, given that Wakefield was absolutely discredited by the academic community and ostracized from the profession.

Carl T. Bergstrom @carlbergstrom.com · 5h

You really don’t get it, do you?

Under the current administration, it would be the people who debunked Wakefield who were put on trial for fraud.

1 1

Carl T. Bergstrom @carlbergstrom.com · 7h

If it had actually happened, yes.

🧵

Andrew @generalising.bsky.social · 1d

I got very curious about this. You can trace Shearer's unit (21 CCS) and you can see what looks like the echo of his little blaze of fame: in Sept 1916 you get visits from various grandees including ... the editor of the BMJ, who presumably came to see for himself.

Carl T. Bergstrom @carlbergstrom.com · 7h

(to be fair, it’s mostly dipshit tech bros who write about this, but it’s shocking to see it so widely endorsed among people who should know better)

4 63

Carl T. Bergstrom @carlbergstrom.com · 7h

It amazes me that academics living under an authoritarian US government that has a various times labeled both climate change and vaccine science as fraud still seem that it would be a good idea to criminalize scientific misconduct.

3 17 180

Reposted by Carl T. Bergstrom, Iñaki Aldasoro

Reuters @reuters.com · 7h

OpenAI to allow mature content on ChatGPT for adult verified users starting December reut.rs/3J8NTAi

OpenAI to allow mature content on ChatGPT for adult verified users starting December

OpenAI will allow mature content for ChatGPT users who verify their age on the platform starting in December, CEO Sam Altman said, after the chatbot was made restrictive for users in mental distress.

reut.rs

12 5 18

Carl T. Bergstrom @carlbergstrom.com · 9h

I wouldn’t have picked an example where the LLM would have gotten 50% on a middle school algebra test for my slam dunk on humans.

3 2 56

Reposted by Carl T. Bergstrom, Janet Murray, Michael D. McDonald , and 3 more Carl T. Bergstrom, Janet Murray, Michael D. McDonald, Kate Starbird, Iñaki Aldasoro, Patrick Stokes

The Verge @theverge.com · 9h

Sam Altman says ChatGPT will soon sext with verified adults

ChatGPT is about to get flirty.

buff.ly

190 28 240

Carl T. Bergstrom @carlbergstrom.com · 21h

🤣 Oh man good point. I should ask Brian to hold my beer more often.

Carl T. Bergstrom @carlbergstrom.com · 22h

"Hey Wansink, can you hold my beer for a sec?"

Brain nods and takes the can.

"I just need to post this clever work hack to Bluesky real quick...."

1 31

Reposted by Carl T. Bergstrom, David Underdown

Andrew @generalising.bsky.social · 1d

2 16 32

Carl T. Bergstrom @carlbergstrom.com · 1d

An excellent response from Ben Harrap.

I can only play in your back yard if you invite me in – Ben Harrap

benharrap.com

1 6 26

Carl T. Bergstrom @carlbergstrom.com · 1d

👏👏👏

Carl T. Bergstrom @carlbergstrom.com · 1d

That instagram page!

Nucieus?

And of course the labels all wrong.

AI art (?) of cell with labels misspelled and in the wrong places.

1 3 14

Carl T. Bergstrom @carlbergstrom.com · 2d

I know everyone’s struggling to justify their investment in AI, but “LinkedIn AI will prevent you from hiring actual vampires” is not the strongest use case one might’ve imagined.

www.linkedin.com/posts/jensen...

Don't hire energy vampires! LinkedIn is the network here to help your small business hire the RIGHT people, fast. Please see our new ad below. Embracing Halloween AND great AI-driven hiring! | J...

Don't hire energy vampires! LinkedIn is the network here to help your small business hire the RIGHT people, fast. Please see our new ad below. Embracing Halloween AND great AI-driven hiring!

www.linkedin.com

4 7 83

Carl T. Bergstrom @carlbergstrom.com · 2d

Never thought I’d miss the fucking Marlboro man

4 4 120

Carl T. Bergstrom @carlbergstrom.com · 2d

I’ve got no beef with people who choose to gamble on sports and can do so responsibly.

What I have a problem with is trying to grow the market—let alone pull people with gambling problems back into the market

1 20

Carl T. Bergstrom @carlbergstrom.com · 2d

Am I just being a Puritan?

I think it’s bad enough that the MLB allows betting advertising in between innings, but putting up an ad with the betting line or whatever in a graphic on screen during play?

Given the prevalence of problem gambling it’s IMO inexcusable.

49 35 550

Carl T. Bergstrom @carlbergstrom.com · 2d

This is so many levels of funny I've got nothing in response.

Carl T. Bergstrom @carlbergstrom.com · 2d

Until they hand shit off to Gemini. Stay tuned....

1 18

Reposted by John R. Hutchinson

Carl T. Bergstrom @carlbergstrom.com · 2d

Left: In a desperate effort to find reasonable use case, OpenAI is running ads during the baseball playoffs about how you can use ChatGPT to plan a road trip. www.youtube.com/watch?v=go4o...

Right: What happens if you actually do that. www.bbc.com/news/article...

Open AI commercial about using ChatGPT to plan a road trip

Hannah Read, 37, tried to use an AI chatbot to plan a trip to Norway with her partner and three children.

She wanted to drive from her home in Flintshire, north Wales and cross the North Sea by ferry.

"I thought it might make for a nice drive," Hannah says. "I asked ChatGPT if there was a ferry from the UK to Norway and it said there was one from Newcastle to Bergen."

She later checked a ferry travel website but found no such route exists.

A travel blog detailing how Brits can reach Norway says the last time the route operated was in 2008.

"I did feel a bit disappointed when I found out the information on ChatGPT was incorrect, as I'd got quite excited and had started planning the trip in my head," Hannah adds.

"My advice is don't rely on AI 100%, it's better to still do proper research."

8 63 260

Carl T. Bergstrom @carlbergstrom.com · 2d

More context in this thread:

bsky.app/profile/mark...

Mark Histed @markhisted.org · 3d

Seeing a lot of celebration of this as a win.

But at NIH we have seen this exact play before.

It looks like it is the Vought strategy -and to stop the lawless firings we have to see the bigger picture.
A 🧵:

Trump Administration Will Rehire Scores of Experts Fired in Error

www.nytimes.com

12 65

Reposted by Meri Koivusalo

Carl T. Bergstrom @carlbergstrom.com · 2d

Thankfully, some of yesterday's CDC firings have been rescinded.

But this should be a major scandal, not business as usual in this shitshow of an administration. It's either spectacular incompetence or an effort to make even politically untouchable positions feel precarious to those in them.

Helen Branswell 🇨🇦 @helenbranswell.bsky.social · 3d

100s of #CDC employees were notified Friday they were being fired — RIF'd. The firings cross multiple parts of the agency, endangering its core mission of keeping Americans safe.
Saturday, some were informed they were RIF'd in error.
Hard to imagine how this happens www.statnews.com/2025/10/11/c...

CDC battered by government shutdown firings, while some are rescinded

The White House’s mass firing of CDC staff on Friday has decimated offices related to injury prevention, respiratory disease surveillance, and chronic disease, according to four people familiar with t...

www.statnews.com

15 210 790