Lightnews — Scholar-powered news

Deb Raji @rajiinio.bsky.social · 12d

There is so much about navigating the Internet in a low resourced language that makes one unnecessarily vulnerable to malicious actors. It's not just a quality of experience difference, but literally the soft belly through which misinformation spreaders attack.

Hellina Hailu Nigatu @hellinanigatu.bsky.social · 13d

Very excited for our upcoming #AIES paper Into the Void: Understanding Online Health Information in Low-Web Data Languages.

Link: arxiv.org/pdf/2509.20245

1/n

arxiv.org

6 10

Deb Raji @rajiinio.bsky.social · 12d

Congrats, Hellina!

1 1

Deb Raji @rajiinio.bsky.social · 26d

More of this kind of reporting, please!

5 31 150

Reposted by Deb Raji

Roberta Fischli @leonieclaude.bsky.social · 26d

So inspiring to see @karenhao.bsky.social in conversation with @rajiinio.bsky.social at the AI and Society event @ Berkeley. We need more formats like this in the Bay Area! (and beyond)

1 4

Deb Raji @rajiinio.bsky.social · Sep 7

😂😂 lol why is this interaction so on brand

1

Deb Raji @rajiinio.bsky.social · Aug 23

I get similar emails as well! Honestly, very strange ...

1 8

Deb Raji @rajiinio.bsky.social · Aug 12

Hm, that's interesting. What's the scientific hypothesis tested with for example Imagenet? The hold out method?

1 1

Deb Raji @rajiinio.bsky.social · Aug 12

I think there's an interesting normative difference revealed in this spat - namely that there's an unfairness to infinite versioning and selective disclosure, which impacts *leaderboard placement* (ie. which models we see on top as number 1) but might still be statistically sound evaluation practice

1 2

Deb Raji @rajiinio.bsky.social · Aug 12

Hm, curious - what are your thoughts on LM Arena's response to the Leaderboard Illusion paper? news.lmarena.ai/our-response/

LMArena Reponse to "The Leaderboard Illusion" Writeup

We’re actively working with the authors to address and amend the factual concerns identified in the paper.

news.lmarena.ai

1 2

Deb Raji @rajiinio.bsky.social · Aug 4

congrats, ben!!

1

Reposted by Deb Raji

Ben Recht @beenwrekt.bsky.social · Aug 4

I’m excited to announce that my new book, _The Irrational Decision_, is now available for pre-order from Princeton University Press.

The Irrational Decision

How the computer revolution shaped our conception of rationality—and why human problems require solutions rooted in human intuition, morality, and judgment

press.princeton.edu

9 30 120

Reposted by Deb Raji

Emma Harvey @emmharv.bsky.social · Jul 24

Towards AI Accountability Infrastructure: Gaps and Opportunities in AI Audit Tooling by @victorojewale.bsky.social @rbsteed.com @briana-v.bsky.social @abeba.bsky.social @rajiinio.bsky.social compares the landscape of AI audit tools (tools.auditing-ai.com) to the actual needs of AI auditors.

Screenshot of paper title and author list:

Towards AI Accountability Infrastructure: Gaps and Opportunities in AI Audit Tooling
Victor Ojewale, Ryan Steed, Briana Vecchione, Abeba Birhane, Inioluwa Deborah Raji

1 4 23

Deb Raji @rajiinio.bsky.social · Jul 26

Emma has such good research taste :)

Given the sheer scale of these events, its really helpful to see what caught people's eye at these conferences...

Emma Harvey @emmharv.bsky.social · Jul 14

After having such a great time at #CHI2025 and #FAccT2025, I wanted to share some of my favorite recent papers here!

I'll aim to post new ones throughout the summer and will tag all the authors I can find on Bsky. Please feel welcome to chime in with thoughts / paper recs / etc.!!

🧵⬇️:

2 9

Reposted by Deb Raji

Emma Harvey @emmharv.bsky.social · Jul 14

After having such a great time at #CHI2025 and #FAccT2025, I wanted to share some of my favorite recent papers here!

I'll aim to post new ones throughout the summer and will tag all the authors I can find on Bsky. Please feel welcome to chime in with thoughts / paper recs / etc.!!

🧵⬇️:

2 10 54

Deb Raji @rajiinio.bsky.social · Jul 24

Was beyond disappointed to see this in the AI Action Plan. Messing with the NIST RMF (which many private & public institutions currently rely on) feels like a cheap shot

2 5 11

Deb Raji @rajiinio.bsky.social · Jul 24

The way this was predictable from the start...

Tombmas Spooks @thomasfuchs.at · Jul 24

The purpose of a system is what it does

Headline about FDA “AI” making up studies for drug approvals

1 3 16

Deb Raji @rajiinio.bsky.social · Jul 22

This group (+ @leonyin.org ) makes this team like the Avengers of data journalists lol

Congrats to Bloomberg!

Dhruv Mehrotra @dmehro.bsky.social · Jul 21

Friday was my last day at WIRED. Today I started a new job at Bloomberg on a dream desk with @suryamattu.com and @jeffykao.bsky.social.

Got a tip? A dataset? Something we should look at? Find me on Signal at dmehro.89

In no particular order, here’s some stuff I’d like to continue to dig in to:

9

Reposted by Deb Raji

Henry Farrell @himself.bsky.social · Jul 15

The genius of the Wired people, as I see it from outside, is that they very quickly saw how a specific model of _tech_ reporting could be much more easily adapted into a closely related model of _political_ reporting, in a world where actual and organizational technology is urgently relevant.

4 29 210

Deb Raji @rajiinio.bsky.social · Jul 13

After my participation in the AI Senate forum, Andreesen went out of his way to find me on X and block me (I had never interacted w him).

Just a terrifyingly hateful person. I remain deeply unsettled the more I learn about how deep his prejudice actually goes.

nitasha tiku @nitasha.bsky.social · Jul 12

New from me. Hope you’ll give it a read www.washingtonpost.com/technology/2...

Tech billionaire Trump adviser Marc Andreesen says universities will ‘pay the price’ for DEI

Tech investor and Trump adviser Marc Andreessen in a private group chat attacked colleges for promoting diversity, saying they had discriminated against “native born kids.”

www.washingtonpost.com

1 4 42

Deb Raji @rajiinio.bsky.social · Jul 9

I've always felt somewhat uncomfortable with the framing of AI risk around the actions of "malicious actors". Because sometimes the malicious actor is the company that built the thing. And the model is causing harm because it was successfully steered into doing what it's creators wanted it to do.

2 11 43

Deb Raji @rajiinio.bsky.social · Jul 9

The distinction between this situation & the 2016 Tay twitter chatbot fiasco by Microsoft feels like a sign of the times - now, it's the companies actively trying to steer the chatbot towards extremist views rather than the interactions of a malicious public pushing the chatbot in that direction...

Charlie Warzel @cwarzel.bsky.social · Jul 8

the 🐐 @matteowong.bsky.social and i tried to explain some of technical reasons for why grok went nazi as well as some of the less technical ones (prompting Grok to use X posts as a primary source and rhetorical inspiration). It's all quite awful and illuminating www.theatlantic.com/technology/a...

Grok, as Musk and xAI have designed it, is fertile ground for showcasing the worst that chatbots have to offer. Musk has made it no secret that he wants his large language model to parrot a specific, anti-woke ideological and rhetorical style that, while not always explicitly racist, is something of a gateway to the fringes. By asking Grok to use X posts as a primary source and rhetorical inspiration, xAI is sending the large language model into a toxic corpus where trolls, political propagandists, and outright racists are some of the loudest voices. Musk himself seems to abhor guardrails generally—except in cases where guardrails help him personally—preferring to hurriedly ship products, rapid unscheduled disassemblies be damned. That may be fine for an uncrewed rocket, but X has hundreds of millions of users aboard.

For all its awfulness, the Grok debacle is also clarifying. It is a look into the beating heart of a platform that appears to be collapsing under the weight of its worst and loudest users. Musk and xAI have designed their chatbot to be a mascot of sorts for X—an anthropomorphic layer that reflects the platform’s ethos. They’ve communicated their values and given it clear instructions. That the machine has read them and responded by turning into a neo-Nazi speaks volumes.

2 7 37

Deb Raji @rajiinio.bsky.social · Jul 6

Instead of single words or phrases, users will immediately attempt multi-hop searches, incorporate literally paragraphs of context, etc lol completely unprecedented search behavior afaik

Interestingly, current benchmarks like SimpleQA don't reflect this & still feature the one liner format

3

Deb Raji @rajiinio.bsky.social · Jul 6

One of the first things I noticed looking through SearchArena logs was the very stark differences between queries for search LLMs (eg. Perplexity AI, etc) and regular search engines.

It really is quite remarkable how different people's expectations are!

Ethan Mollick @emollick.bsky.social · Jul 6

An example of the type of search (would require reading multiple sites, balancing multiple constraints) where o3/Gemini 2.5 Pro has completely replaced Google for me.

1 8

Reposted by Deb Raji

jessica dai @jessica.bsky.social · Jul 1

individual reporting for post-deployment evals — a little manifesto (& new preprints!)

tldr: end users have unique insights about how deployed systems are failing; we should figure out how to translate their experiences into formal evaluations of those systems.

1 3 18

Reposted by Deb Raji

Will Oremus @willoremus.com · Jul 1

In a stunning reversal, the Senate voted 99-1 this morning to strip from Trump's big bill a 10-year moratorium on state-level AI regulations.

Gift link to my story on how it happened and who's celebrating: wapo.st/3TOyiaG

In dramatic reversal, Senate votes to kill AI-law moratorium

A GOP-led bid to stop states from regulating AI collapsed after a deal to save it fell through.

www.washingtonpost.com

19 170 680