Lightnews — Scholar-powered news

Mark Madsen @markmadsen.bsky.social · 18h

Don’t we have police to arrest people who break the law like that?

Reposted by Mark Madsen

Jesse "Pecke in the Crowne" Locker @jessemlocker.bsky.social · 19h

Armed federal agents want you to believe they were assaulted by a mom dressed as a banana playing the Ghostbusters theme on the clarinet

Federal agents barrel into band at Portland ICE protest, arrest clarinetist, accuse her of assault

Oriana Korol, a 38-year-old child and family therapist, remains in Clark County Jail.

www.oregonlive.com

1 19 40

Reposted by Mark Madsen

Aubrey Hirsch @aubreyhirsch.bsky.social · 1d

Did you know that if you buy my book on Amazon, Jeff Bezos makes 5 times as much as I do on each copy sold? Sharing some behind-the-scenes book math on why buying directly from presses is best for EVERYONE (except Bezos) in my newsletter today! aubreyhirsch.substack.com/p/book-math

27 460 820

Reposted by Mark Madsen

Ghostly Bree @mostlybree.kitrocha.com · 19h

Canceling most of my national subs and redirecting them to local news, independent journalists, and places writing about things I care about was my best financial choice of the year. Yay past Bree.

1 4 45

Mark Madsen @markmadsen.bsky.social · 18h

For my UK friends, Mike Johnson is what you call a bellend.

The Urbanist @theurbanist.org · 19h

Dear Mike Johnson...

1

Mark Madsen @markmadsen.bsky.social · 18h

I must have good blocklists because I never see these posts people refer too.
Grateful to the maintainers

4

Reposted by Mark Madsen

Karen Hao @karenhao.bsky.social · 19h

At the TIME100 AI celebration last night, Natasha Lyonne stood in front of the emperors of AI and challenged them to their face. Doing that takes guts. And I’m incredibly honored that she cited my work as well.

Thank you for speaking truth to power, Natasha. 🙏

youtube.com/watch?v=RUu9...

Natasha Lyonne Asks AI Leaders: "Whatever Happened to Being Cool?"

YouTube video by TIME

m.youtube.com

3 23 100

Mark Madsen @markmadsen.bsky.social · 18h

This isn’t a new pricing experience.
When relational databases went to the cloud/aaS model they had the same problem and had to use resource cost or a fiction (credits) that mapped to resources.
This was true in mainframe timeshare days too.
The answer then was to run your own servers.

Ed Zitron @edzitron.com · 19h

There's one person who pays them $250 a month who cost Augment $15,000. Amazing stuff. Can only imagine how bad things are for Cursor or Replit
www.augmentcode.com/blog/augment...

The user message model also isn’t sustainable for Augment Code as a business. For example, over the last 30 days, a user on our $250 Max plan has issued 335 requests per hour, every hour, for 30 days, and is approaching $15,000 per month in cost to Augment Code. This sort of use isn’t inherently bad, but as a business, we have to price our service in accordance with our costs.

1

Mark Madsen @markmadsen.bsky.social · 18h

What now? I’ve been deferring updates so I haven’t experienced the horror

Reposted by Mark Madsen

Rebecca Solnit @rebeccasolnit.bsky.social · 19h

In other words:

404 Media @404media.co · 21h

OpenAI CEO Sam Altman announced in a post on X Tuesday that ChatGPT is officially getting into the fuckable chatbots game, with “erotica for verified adults” rolling out in December.

🔗 www.404media.co/chatgpt-erot...

ChatGPT’s Hail Mary: Chatbots You Can Fuck

As recent reports show OpenAI bleeding cash, and on the heels of accusations that ChatGPT caused teens and adults alike to harm themselves and others, CEO Sam Altman announced that you can soon fuck t...

www.404media.co

10 17 110

Reposted by Mark Madsen

spookillumi @illumi.meme · 22d

me telling my grandkids what it was like to have vaccines

The Simpsons old man sitting on a stump telling a story but all the kids have been replaced with headstones

42 2K 8K

Reposted by Mark Madsen

Octopus/Caveman @octopuscaveman.bsky.social · 13d

This just feels mean lol

2 3 27

Mark Madsen @markmadsen.bsky.social · 21h

I saw that picture and my mind immediately went to this picture of him at his Surrey home two decades later

Heavyset hairy man in a hot tub talking on two phones, surrounded by telephones, a TV, pictures, a gold clock, trophies, and an american flag

1

Mark Madsen @markmadsen.bsky.social · 22h

OpenAI's models are amazing. They can replace your radiologist with a bot that cheerfully misdiagnoses and congratulates you, then offers to send you hyper-realistic dick pics when you get home.

Dr Abeba Birhane @abeba.bsky.social · 22h

i don't have the words to emphasise how seriously bad things are gonna get

8 46

Reposted by Mark Madsen

good hyuck, babe! @markpopham.bsky.social · 22h

if pep boys said that now the pep boys could call you up and talk sexy to you you wouldn't think that pep boys was doing really well, as a company

12 110 510

Reposted by Mark Madsen

Hypervisible @hypervisible.blacksky.app · 22h

Definitely nothing can go wrong with OpenAI having a full record of your sexy chatbot interactions.

13 85 350

Reposted by Mark Madsen

Ygrene @ygrene.bsky.social · 3d

reminder to check your kid’s trick or treat candy this year to make sure there isn’t any ai in it

11 130 520

Reposted by Mark Madsen

Drew Harwell @drewharwell.com · 1d

Sam Altman went from “AI will cure cancer" to "ChatGPT porn" in less than a month

ChatGPT boss predicts when AI could cure cancer

We made ChatGPT pretty restrictive to make sure we were being careful with mental health issues. We realize this made it less useful/enjoyable to many users who had no mental health problems, but given the seriousness of the issue we wanted to get this right.

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.

In a few weeks, we plan to put out a new version of ChatGPT that allows people to have a personality that behaves more like what people liked about 4o (we hope it will be better!). If you want your ChatGPT to respond in a very human-like way, or use a ton of emoji, or act like a friend, ChatGPT should do it (but only if you want it, not because we are usage-maxxing).

In December, as we roll out age-gating more fully and as part of our “treat adult users like adults” principle, we will allow even more, like erotica for verified adults.
12:02 PM · Oct 14, 2025
·
1.2M
Views

130 860 3.2K

Reposted by Mark Madsen

karl rove knausgård @uhactually.bsky.social · 1d

You might think Ryan is exaggerating but I used to go to a regular dinner that gradually got taken over by young Republican staffers and I promise you he is not.

ryan cooper @ryanlcooper.com · 1d

every young Republican staffer is a Nazi, it's that simple www.politico.com/news/2025/10...

‘I love Hitler’: Leaked messages expose Young Republicans’ racist chat

Thousands of private messages reveal young GOP leaders joking about gas chambers, slavery and rape.

www.politico.com

13 300 1.3K

Reposted by Mark Madsen

John @notheotherjohn.bsky.social · 1d

We're finally getting rid of "woke" lead-free protein drinks

paris martineau @paris.nyc · 1d

my latest investigation for @consumerreports.org is based on months of reporting and 60+ lab tests of leading protein supplements

we found that most protein powders and shakes have more lead in one serving than our experts say is safe to have in a day (🧵)

www.consumerreports.org/lead/protein...

Protein Powders and Shakes Contain High Levels of Lead - Consumer Reports

CR tests of 23 popular protein powders and shakes found that most contain high levels of lead.

www.consumerreports.org

1 1 3

Reposted by Mark Madsen

Alex Degen (Magic Serple 2025) @adactivity.bsky.social · 1d

This is cool after the war on everything called “porn” by platforms and payment processors. A machine that stole all your words can generate and sell “erotica” but you can’t haaa

Matt Burgess (WIRED) @mattburgess1.bsky.social · 1d

Oh good, ChatGPT is getting "erotica for verified adults" later this year

37 2.4K 5.2K

Mark Madsen @markmadsen.bsky.social · 1d

I had not noticed that people are talking about this again. I guess bit by bit we reconstruct the past that led up to 1992.
Soon we will rediscover Cyc and commonsense reasoning!

1

Mark Madsen @markmadsen.bsky.social · 1d

I have a bit about world models in a presentation I'm working on.
Mainly because it's about language as interface and the problem of also having UI controls that can change state. If the AI lacks a world model or ability to evaluate state this becomes a problem. And we're back at cause & effect

1

Mark Madsen @markmadsen.bsky.social · 1d

This is an example of what happens when there is no competence. Things moving in opposite directions are moving toward each other. Reasoning model works great!
Meanwhile, companies are putting applications using this model into HR to score job applicants and determine who gets raises.

An example reasoning model session:
The question is about a man and an apple moving in opposite directions at different speeds: how long until they meet?
The "facts" as determined by the model are:
man and apple are 187 km apart
Man at 50 kph
Apple at 20 kph
They are traveling toward each other (since they are on opposite sides of the river and moving in opposite directions)

This is state of the art reasoning. We were doing better in 1992.

Mark Madsen @markmadsen.bsky.social · 1d

Therefore, always make sure that evaluations of AI (in papers, or in your own work) test for both performance and competence.
If you evaluate on performance alone you will have surprising failures after it's been deployed.

A drive through pharmacy where a car drove through the wall of the pharmacy. Probably a Tesla driving system

1