Lightnews — Scholar-powered news

Mark Riedl @markriedl.bsky.social · 6h

This is like the wholesome cozy version of The Onion.

Anyway I have a new hero

Unofficial BBCNews (World) Bot @bbcnews-world-rss.bsky.social · 13h

Scientist on three-week off-grid hike finds out he's won the Nobel prize

Dr Fred Ramsdell was "living his best life" offline when the Nobel committee tried to contact him.

www.bbc.com

1 14

Mark Riedl @markriedl.bsky.social · 16h

Hi all,

Georgia Tech is holding its 2nd Summit on Responsible Computing, AI, and Society October 28-29th in Atlanta.

rcais.github.io

Our speakers and schedule is set. It's going to be great!

If you are planning to be in Atlanta and want to join us, registration is now open on the website.

7 18

Reposted by Mark Riedl

Bookshop.org @bookshop.org · 1d

Can’t decide what to buy on Prime Day?

Try: absolutely nothing, and then go support indie bookstores instead 📚

62 2.7K 7.5K

Mark Riedl @markriedl.bsky.social · 22h

Oh uh, am I cliche? bsky.app/profile/mark...

Mark Riedl @markriedl.bsky.social · 1d

If you're at #COLM2025 on Friday Oct 10, I'll be giving an invited talk in the Workshop on the Application of LLM Explainability to Reasoning and Planning.

The title is: "Reasoning, Explanation, and Human-Centered AI".

Below I have tentatively diagrammed the flow of my talk.

1 3

Reposted by Mark Riedl

David Mimno @dmimno.bsky.social · 1d

COLM word cloud. Yoav says it’s the year of reasoning, but evaluation is also huge.

Evaluation reasoning interpretability rl in context benchmark alignment synthetic data

3 5 27

Reposted by Mark Riedl

Maria Antoniak @mariaa.bsky.social · 1d

Here’s a #COLM2025 feed!

Pin it 📌 to follow along with the conference this week!

2 17 24

Reposted by Mark Riedl

Dr Kat Day (she/her) @chronicleflask.katday.com · 1d

Daughter of Robin Williams, Zelda Williams 💔

zeldawilliams
20m
From Create Mode
••
Please, just stop sending me AI videos of Dad. Stop believing I wanna see it or that I'll understand, I don't and I won't. If you're just trying to troll me, I've seen way worse, i'll restrict and move on. But please, if you've got any decency, just stop doing this to him and to me, to everyone even, full stop. It's dumb, it's a waste of time and energy, and believe me, it's NOT what he'd want.
To watch the legacies of real people be condensed down to 'this vaguely looks and sounds like them so that's enough,
, just so
other people can churn out horrible TikTok slop puppeteering them is maddening. You're not making art, you're making disgusting, over-processed hotdogs out of the lives of human beings, out of the history of art and music, and then shoving them down someone else's throat hoping they'll give you a little thumbs up and like it.
Gross.

And for the love of EVERYTHING, stop calling it 'the future. Al is just badly recycling and regurgitating the past to be reconsumed. You are taking in the Human Centipede of content, and from the very very end of the line, all while the folks at the front laugh and laugh, consume and consume.

79 6.9K 16K

Reposted by Mark Riedl

Max Woolf @minimaxir.bsky.social · 2d

accidentally caused Nano Banana's system prompt to leak, but in a fun way

6 11 84

Mark Riedl @markriedl.bsky.social · 1d

If you're at #COLM2025 on Friday Oct 10, I'll be giving an invited talk in the Workshop on the Application of LLM Explainability to Reasoning and Planning.

The title is: "Reasoning, Explanation, and Human-Centered AI".

Below I have tentatively diagrammed the flow of my talk.

17

Mark Riedl @markriedl.bsky.social · 2d

Finally had to upgrade my iPhone to the version with the Liquid Glass UI. I appreciate the Apple UI team’s gradual evolution from “too many notifications” to “too many notifications but now you can’t read them”

3 28

Mark Riedl @markriedl.bsky.social · 2d

#COLM2025 is starting. Find @glennmatlin.bsky.social to learn about why LLMs kinda, sorta, but don't really understand psychological suspense in stories.

Mark Riedl @markriedl.bsky.social · Aug 25

Our work on Detecting Suspense in Stories will appear at COLM'25 arxiv.org/abs/2508.15794

Do LLMs know when stories are suspenseful?

We ran LLMs through a bunch of classical psychology suspense studies

The answer: kinda, sorta? Honestly, better than I expected. But the results are nuanced...

1/

2 1 12

Mark Riedl @markriedl.bsky.social · 2d

I evaporated some water in order to bring you this video of evaporating water sora.chatgpt.com/p/s_68e30039...

notalldystopias on Sora

Steam pouring out of several evapco rooftop evaporators, long shot from above

sora.chatgpt.com

1 2

Mark Riedl @markriedl.bsky.social · 2d

My local bookstore is amazing littleshopofstories.com

Live D&D session featuring local authors @runwithskizzers.bsky.social @jacksonpearce.bsky.social @markdoesstuff.bsky.social. They were great. Like having Critical Role in one’s backyard

1 2 20

Mark Riedl @markriedl.bsky.social · 3d

Interesting point

Ben Recht @beenwrekt.bsky.social · 3d

In non-LLM AI news, Waymo has quietly been involved in two fatal car crashes in 2025. Phil Koopman details that this fatality rate is now comparable to humans.

People: 7/billion miles in all conditions
Waymo: 5/billion miles in geofenced conditions

Blame-Free Robotaxi Crashes Are Still Crashes

The tricky logic behind robotaxi crash metrics

philkoopman.substack.com

2 6 22

Mark Riedl @markriedl.bsky.social · 4d

AI is going to take over the wo—oh are you stressed out little buddy? Here, let me help you

dan @danabra.mov · 4d

apparently when Claude starts lying and trying to evade work, it means it's stressed out and you need to reduce its level of stress. e.g. scope down the task a bit more, suggest to break it down into parts, suggest to extract a minimal repro

1 1 37

Mark Riedl @markriedl.bsky.social · 4d

Well that was quite the rabbit hole. tfwiki.net/wiki/Reprodu...

1 2

Mark Riedl @markriedl.bsky.social · 5d

Optimus Prime is a robot that turns into a semi truck. I do not want to have to think about whether they have small or large reproductive cells.

1 4

Mark Riedl @markriedl.bsky.social · 5d

Drone delivery is going well

Gizmodo @gizmodo.com · 5d

Amazon Delivery Drones Crash Into a Crane in Arizona, Feds Investigating https://gizmodo.com/amazon-delivery-drones-crash-into-a-crane-in-arizona-feds-investigating-2000667298

Amazon Delivery Drones Crash Into a Crane in Arizona, Feds Investigating

Regulators have launched probes after two Prime Air drones collided with a construction crane.

gizmodo.com

1 5

Mark Riedl @markriedl.bsky.social · 5d

Overheard in the wild: a middle-schooler calling someone else “AI” to accuse them of making things up.

2 9 37

Mark Riedl @markriedl.bsky.social · 6d

I can’t believe Apple is surprised by this. Meta and Snap have been talking and showing glasses for years

Gizmodo @gizmodo.com · 6d

Apple Says ‘Oh Shit’ and Reportedly Pivots to Chasing Meta’s Smart Glasses https://gizmodo.com/apple-says-oh-shit-and-reportedly-pivots-to-chasing-metas-smart-glasses-2000666643

Apple Says ‘Oh Shit’ and Reportedly Pivots to Chasing Meta’s Smart Glasses

Sorry, Vision Pro lovers, looks like smart glasses are sucking up all the air.

gizmodo.com

2 4

Mark Riedl @markriedl.bsky.social · 6d

I once sat on an airplane next to someone doing "responsible AI" online training. I couldn't help look at their laptop screen. Nothing was wrong per se, but it was some amazingly vacuous bullet points. I was screaming inside the entire time.

17

Mark Riedl @markriedl.bsky.social · 6d

I would love to see the AI skill training that Accenture used. Anyone interested in leaking it to me?
www.cnbc.com/2025/09/26/a...

Accenture plans on 'exiting' staff who can't be reskilled on AI amid restructuring strategy

Accenture CEO Julie Sweet said as advanced AI becomes core to the company's strategy, employees are expected to "retrain and retool" at scale.

www.cnbc.com

1 2 6

Reposted by Mark Riedl

Ruha Benjamin @ruha9.bsky.social · 7d

Yall please go watch @merriam-webster.com’s new AI ad😅

youtu.be/RvQnl3O0D3c?...

LLM

YouTube video by Merriam-Webster

youtu.be

4 76 160

Mark Riedl @markriedl.bsky.social · 7d

I'm not a vision or video generation expert, but why could sora2 not be so overparameterized that it memorized a chunk of the game?

1

Reposted by Mark Riedl

Drew Harwell @drewharwell.com · 7d

OpenAI employees are very excited about how well their new AI tool can create fake videos of people doing crimes and have definitely thought through all the implications of this

220 3.3K 11K