Excited to share #ICML2026 paper from my internship @ Google DeepMind!
AI models are deployed globally, but AI safety datasets are largely geographically homogenous. What is the impact of culture on AI safety ratings? Is there any impact beyond standard demographics like age, gender, and ethnicity?
Online Link : timesofindia.indiatimes.com/blogs/toi-ed...
Video
timesofindia.indiatimes.com
One would think that an editor at a publishing or media house would be able to tell the difference between human-written and AI-generated text with high precision, the way a seasoned jeweller appraise...
Had a great time speaking with @bcmerchant.bsky.social and @vauhinivara.bsky.social about the current state of AI –– how Anthropic convinced the media that it's the industry's "ethical" leader, and how AI writing is creeping into journalism and literature: www.wnycstudios.org/podcasts/otm...
I’m sorry what??? Great reporting from @knibbs.bsky.social on the author who used AI to write his book about AI and the future of truth. 🤯 www.wired.com/story/future...
This professor of something computer science-y appreciates it :-)
This week, Anthropic filed for an IPO following a valuation of nearly $1 trillion, which would make it one of the largest IPOs in history. On this week’s On the Media, the company’s m...
www.wnycstudios.org
Excellent essay and Excellent writing !!
Cool 😎
Come for @bcmerchant.bsky.social's take on how Anthropic infiltrated the Vatican (and subscribe to his Substack to read about it, while you're at it), and stay to hear me talk about AI turning up in books and magazines!
Micah Loewinger
For @timesofindianews.bsky.social I wrote about why we desperately need to rely on accurate AI detectors to prevent this slopocalypse.
Yes @pangram.com is the best AI detector we’ve got and blanket statements like detectors don’t work are not backed by empirical evidence but isolated anecdotes
Tuhin Chakrabarty
Tuhin Chakrabarty
Tuhin Chakrabarty
Vauhini Vara
Tuhin Chakrabarty
Vauhini Vara
I thought I was writing an essay explaining a decision to stop using the em dash — or least use it way less — because of how AI chatbots have appropriated it. Instead, I found myself arguing passionately in favor of it. 😑
Ran some 🧪 to 🔬 why the Granta short story was certainly 🤖 generated.
A lot of bad writing happens because AI hasn’t learned aesthetics. It has simply memorized the whole internet and called it a day.
So sure, maybe you don't trust AI detectors. But you can trust your own eyes. #AISlop
📻 Show’s up! Featuring @bcmerchant.bsky.social, @vauhinivara.bsky.social, and David Garrett on-the-media.visitlink.me/mMfPo1
If you liked Bears Will Be Boys, you might enjoy our new FAccT paper!
We prompted LLMs to complete 24K stories about animal characters where gender is unstated.
We found that.. bears are *still* boys. And female animal characters disappeared while "neutrality" increased.
arxiv.org/abs/2606.079...
Vauhini Vara
Tuhin Chakrabarty
On the Media
Did the administration successfully bury their horrific change to green card procedures by releasing it Friday afternoon before Memorial Day? People who have followed every rule should not have their lives upturned for months or years. Where is the continued pressure?
