Leaving aside all the ethics of AI fashion models... did no-one look at this before printing??? Just try to work out where her knees are and how they're connected to her feet and hips. Bizarre to choose a zigzag dress that's just going to emphasise any weird contours too

word of the year pls

I had some fun speaking at #sxswlondon this week. Video for FIVE THINGS YOU NEED TO KNOW ABOUT AI should be coming soon!

"We will look back at this period in history and realise that letting a billionaire’s ego drive innovation — rather than as part of a collective effort, as seen with Apollo and the Saturn V — was a grave mistake." www.planetearthandbeyond.co/p/starship-d...

I think I forgot to post this in the travel chaos, but I spoke to @willdouglasheaven.bsky.social for MIT Tech Review about AI, creativity and more, as part of a big feature on how AI is affecting our creativity. I think we have no idea, basically. www.technologyreview.com/2025/04/10/1...

you need to watch these clips

one of the highlights in the issue is this from @willdouglasheaven.bsky.social on how AI can be co-opted to help supercharge creativity in unusual ways. I want to go to an algorave! www.technologyreview.com/2025/04/10/1...

Good breakdown on tariffs

My *third* time on @scifri.bsky.social but first time recording in the NYC studio! Hear me and @willdouglasheaven.bsky.social bring AGI conversations to reality and talk about the issues that really matter, like who controls the power behind AI.

I think Kevin Roose needs to talk to more people outside San Francisco www.nytimes.com/2025/03/14/t...

Possibly the first case of chatbot logs being released under freedom of information laws www.newscientist.com/article/2472...

Yes yes yes! This is hands down one of the best things I've read about AGI, the techno-fetish that's once again grabbing mainstream attention

AGI is whatever we made it to be

Israel is training a spy chatbot on vast numbers of surveilled Arabic calls www.theguardian.com/world/2025/m...

The Revolution will be AI-icized

If you're curious about Muse, Microsoft's new generative gameplay model that produces fuzzy footage of video games that don't (yet) exist, this from @mtrc.bsky.social is easily the best explainer: www.possibilityspace.org/blog-before-...

The act of coding, like sketching or rubber ducking, is also an act of thinking through something. Prompting an LLM might also be a thinking process, but we have to reflect on what each approach helps us with. I think differently while taking notes than just turning it over in my head.

Hi, I spoke to actor Sarah Elmaleh about Harrison Ford’s recent comments on AI and how they’ve brought attention to the current SAG-AFTRA strike www.theguardian.com/games/2025/f...

Blech

OpenAI has put out new guidelines for its models. The updated spec now "explicitly embraces intellectual freedom no matter how challenging or controversial a topic may be." model-spec.openai.com/2025-02-12.h...

“Waste is anything that the person who’s speaking thinks is not a good use of government money.” Really good explainer: @jamesodonnell.bsky.social talks to the people who wrote the spending guide that Musk is now misinterpreting. www.technologyreview.com/2025/02/07/1...

The Chinese AI company DeepSeek surprised observers with its efficient and low-cost model, DeepSeek-R1. @willdouglasheaven.bsky.social, editor for AI @technologyreview.com‬, joins us to discuss the new DeepSeek systems and what might lie ahead for AI.

Yes! slate.com/technology/2...

Anyone out there with knowledge about how software gets installed across US govt agencies (process, authorization etc) free to answer a couple of questions? DM or signal: strawbilly.69

A teen DOGE staffer recently given access to government systems worked at a startup known for hiring convicted hackers. Someone using a Telegram handle associated with him also solicited a cyberattack-for-hire service in 2022. All raising questions about his vetting. www.wired.com/story/edward...

I get to work with @jesshamzelou.bsky.social every day www.technologyreview.com/2025/02/06/1...

"What began as a sprint has become a high-stakes marathon—China’s AI ambitions have never been higher." More good stuff from @caiwei.bsky.social www.technologyreview.com/2025/02/04/1...

Several accounts from USAID workers have now mentioned Gemini being used to surveil them. Here's NY Magazine nymag.com/intelligence...

USAID uses Google, and has for years, so this Gemini install might be coincidental. Lots of silly updates that get pushed through go unnoticed. Not sure one way or the other, but definitely not concrete evidence of anything.

Terrific step-by-step piece on *how* we know that vaccines don’t cause autism www.statnews.com/2025/02/03/v...

Yeah, this is good: "Trump knows the power of marketing. If you make people believe something is true, you make it likelier that it becomes true. Trump has never wanted to be president. He has always wanted to be a king. His plan this time is to first play king on TV. Don't believe him."

NEW: Yes, everyone knows DeepSeek is heavily censored. But we investigated how the censorship actually works on the application level & the post-training level. Understanding that can lead us to ways of "uncensoring" Chinese open-source AIs and making them more useful. www.wired.com/story/deepse...

DeepSeek has pulled back the curtain to reveal that reasoning models are a lot easier to build than people thought. Now things get interesting. New from @willdouglasheaven.bsky.social www.technologyreview.com/2025/01/31/1...

Good stuff lareviewofbooks.org/article/life...

And yet there's more to the DeepSeek story. Building a very efficient model that uses few chips probably requires building one (or more) inefficient models that use a lot of chips first. TLDR don't bet against Big Chip yet This is good blog.heim.xyz/deepseek-wha...

nothing, according to premarket this morning

“The rapid evolution of AI demands agility from Chinese firms to survive.” www.technologyreview.com/2025/01/24/1... Good DeepSeek explainer from @caiwei.bsky.social

“There’s this cartoon called The Flintstones where they have these cars, but they only move when the drivers use their feet. This is sort of how I feel most people are doing AI for software systems.” www.technologyreview.com/2025/01/20/1...

Come and ask us anything you want at noon ET! www.reddit.com/r/technology...

If a dataviz nerd and physics nerd had a baby it would look like this www.economist.com/interactive/...

This summer I took a big swing at a simple question: What exactly are people talking about when they talk about AI? www.technologyreview.com/2024/07/10/1...

MIT Technology Review is offering a whopping 50% off the cost of an annual subscription! 🤠 www.technologyreview.com/subscribe?it...

I love fun stories like this about what happens when you let bots run wild www.technologyreview.com/2024/11/27/1...

"Skeeting is a terrible word for this," he bleeted

Does the world need more books tho www.thebookseller.com/news/new-pub...

Cherry picked or not this tech has got so good so fast runwayml.com/research/int... (but show us the hands)

Really terrific deep dive into Clear, the ubiquitous airport queue-jumping system, and its ambitious plans to expand its service *everywhere* where identity is an issue. Great stuff from @eileenguo.bsky.social as ever. www.technologyreview.com/2024/11/20/1...