AI

The biggest AI conference in the world just embedded hidden prompt injections into its own papers to catch its own reviewers cheating.

AI is changing the world, giving young people terminal brain rot, and stealing our jobs, but is it also trying to kill us? Daniel Kokotajlo, a former OpenAI ...

By: Decca Muldowney and Emily M. Bender Emily was interviewed for two documentaries which premiered at Sundance this year: The AI Doc: How I Became an...

Palantir holds over £670M in UK public contracts across NHS, MoD, police, and now the FCA. Two MoD engineers call it a national security threat. Here is how it happened.

AI will (already has?) killed the inbound job application. If you're a graduating senior looking for a job, here's what you should do: Step 1: Build a presence on the web (portfolio website) 1. Get a website with a domain (preferably your name). 2. Use AI to build out a portfolio website. You want to show what you've done, don't just tell. The more photos/videos you have of projects the better. 3. Host your website on a Git repo and deploy via @cloudflare pages (for free). While you're at it, setup Cloudflare Email Routing to forward email from a nice clean address on your custom domain to an email you already own. Even better, setup a cheap email using @spaceship. Step 2: Talk to people 4. Think about the companies you want to work for and the kind of skills that you could offer to their business. 5. Find alums/friends/friends-of-friends who are doing something that you find interesting in a field where you've built skills and contribute. 6. Write a short and sweet email (don't use AI for this!) to ask if you could hop on a short phone/Zoom call with a few of these people in order to learn more about their journey/career path. (NB: You are not asking for a job here, you're just looking to learn about their journey). You'll get lots of insights from even a few of these calls. 7. At the end of the call (or if they decline a call) ask if there is anyone else that they would be willing to connect you with as a next conversation. It will be clear to everyone that you are looking for a job (probably will come up in conversation naturally) but you don't need to lead with that. Whatever you do, don't put that LinkedIn "Open To Work" badge on your profile photo. Alums especially will know where you've been (they've been there themselves) and will want to help you. Nine times out of ten you won't have to explicitly ask, someone will ask you!

ZwillGen's Jey Kumarasamy writes AI agents introduce meaningfully different risks than traditional AI applications, but strategies can be implemented to manage those risks.

The Slow AI Curriculum for Critical Literacy: 2026 Handbook Status. This is a living handbook. Date. 3 March 2026. Version. 1.2 CPD Accreditation. This programme is accredited by The CPD Group. 25 CPD credits. Course #1019972. Verify at thecpdregister.com. Foundational Orientation The Slow AI ...

Anna Mills asked me about friction and AI.

The dominant model for AI literacy was discredited in science communication forty years ago. Here is what the evidence says works instead.

Attached: 1 video You don’t need to understand the math behind large language models in order to get that they are not minds.

Bruce Maxwell, professor of computer science at Northeastern University, was grading exams for his online master’s course in computer vision, a subfield in artificial intelligence that deals with images, when...

Why the Age of AI Demands We Take Character Formation Seriously

The economic shocks may well be coming, but we have already entered an age of profound uncertainty about ourselves and the world around us.

A newly published study of how college students interact with chatbots and human strangers showed talking to a random person offers more connection than an LLM.

How to set up Claude the right way (so you actually stop going back to ChatGPT).

How to set up Claude Cowork (once you're ready to level up from ChatGPT):

Stop typing. Starting talking to your computer.

Here is a candid 2016 photo of me contributing to an AI training dataset, captured during an entirely ordinary family moment. Except that it was not framed that way then, and it certainly was not understood that way. My children and I had travelled to a capital city to explore, to learn history, and, like millions of others in 2016, to enjoy the childish novelty of chasing rare Pokémon through real streets and public spaces. Instead, we ended up confined to a hotel room for three days with 40 degree fevers, and somewhere in that blur of exhaustion and distraction, a Zubat magically appeared above my bed and my children, delighted, took a photo. It was a small, innocent moment. A sick family, excited children, a fleeting distraction that made them laugh. I did not take the photo. I did not upload it. I did not agree to anything beyond playing a children’s game that appeared, on the surface, to be exactly what it claimed to be. Ten years later, we are told a different story. Niantic just disclosed that photos and AR scans collected through Pokémon Go have produced a dataset of over 30 billion real-world images. The company is now using that data to power visual navigation AI for delivery robots. The 2016 moment, along with billions of others, was not just gameplay. It was data capture. It was long term storage. It was aggregation at global scale. And now, it is being repurposed and monetised as training material for systems designed to navigate and interpret the physical world, including robotics. That is a consent shift. The question that follows 10 years later is not technical. It is ethical, legal, and increasingly unavoidable. What exactly did we consent to? Because consent is not a line buried in terms and conditions that no reasonable person reads, nor is it a blanket permission that stretches indefinitely into future use cases that did not exist, were not explained, and could not have been meaningfully understood at the time. This photo was taken in the European Union, a jurisdiction that has built its identity around the protection of personal data and fundamental rights. Yet here we are, a decade later, confronting the reality that data generated in moments of ordinary life may now sit inside systems far removed from their original context, used in ways that were neither transparent nor foreseeable. We are not simply dealing with innovation outpacing regulation. We are confronting a model in which human experience is quietly converted into long term, transferable, and monetisable infrastructure. The children thought they were catching a 2016 Pokémon. What they were actually doing was helping to train 2026 robots. And no one stopped to inform their parent of this, or ask for consent. Grateful that I was properly dressed… | 44 comments on LinkedIn

I think we might be kidding ourselves about putting humans in the loop in educational contexts.

3.5 million books were published last year. That's too many.

The results from a new March 2026 NBC News poll of 1,000 registered voters showed the only two items on their list with lower negative net ratings than AI’s 26 percent positive / 46 percent negative split were Democrats and Iran. About 56 percent said they had used an AI platform like ChatGPT or Copilot in the previous month. [Image: https://platform.theverge.com/wp-content/uploads/sites/2/2026/03/nbc-news-ai-poll.jpg?quality=90&strip=all]

For designers, who will choose for themselves and for the rest of us.  💡Nerd Rating: 3/5. Remarks prepared for the "From Interface to Agency: A New Discourse for Design and AI" forum on design practices for emerging technology at UC Berkeley, which this year focused on Design for Agentic

I suspect that, over time, many people will end up with some form of NAME.md that travels with them between tools. It will sit next to CVs and portfolios as part of how we present ourselves in digital work.

It’s ‘Tool Time’ again! This month, I want to share some new AI slide tools. And if you haven’t been following, some of these are getting pretty good. I’m not ready to…

The most powerful way to use AI is as a collaborative mind you design intentional systems around starting with a clear sense of what you're optimizing for.

What is the future of work? How is your organization preparing for the future of work? “The future is already here – it's just not very evenly distributed" - William Gibson The fifth World Economic Forum Future of Jobs Report is, like its predecessors, an absolute treasure trove of insightful analysis. Three standout highlights are: ✴️ By 2030, 170 million new jobs will emerge, yet 92 million will be displaced—a net increase of 78 million roles m ✴️ Skill gaps are the biggest barrier to business transformation with 63% of employers identifying them as a major barrier over the 2025- 2030 period. ✴️ On average, workers can expect that two-fifths (39%) of their existing skill sets will be transformed or become outdated over the 2025-2030 period. A recent report from McLean & Company identified 6 impacts organizations can focus on now to effectively respond to the “future of work tsunami” 1️⃣ Mission,vision,values:A value proposition for all. To attract and retain top talent in the future, organizations will need to revisit their value proposition for not only their employees, but also the contingent workforce and society at large. 2️⃣ Employee development: Durable over technical. Development is shifting to become more personalized and continuous. To ensure this evolution produces a workforce equipped with the skills necessary for the future of work, organizations must ensure employees are bought into the need for development, especially considering that survey data indicates that only 46% of employees reported the need for skillsets to change and adapt to the future of work compared to 61% of leaders. 3️⃣ Organizational design: Fluidity and flexibility. A strong understanding of skills, enabled by technology, will be a competitive differentiator for organizations in the future of work. 4️⃣ Leadership: Managing energy, not tasks. The definition of what it means to be a leader is undergoing a profound transformation. Most people surveyed (73% of leadership respondents and 58% of employee-level respondents) feel that leadership skill sets will need to change completely or almost completely to adapt to the future of work in 2030. 5️⃣ Collaboration: No longer exclusively human. With leadership respondents being 1.4 times more likely to anticipate technology having a positive impact on collaboration compared to employees, organizations must optimize collaboration for the future workforce. 6️⃣ Wellbeing: Surviving the loneliness epidemic. While advancements in technology present exciting opportunities for productivity, severe risks associated with isolation and loneliness threaten worker wellbeing in the future of work Acknowledgments: Future of Work Research team at McLean & Company. Future of Jobs Authors: Attilio Di Battista, Sam Grayling, Ximena Játiva, Till Alexander Leopold, Ricky LI, Shuvasish Sharma, and Saadia Zahidi. Original infographic by Jeroen Kraaijenbrink | 72 comments on LinkedIn

Four evidence-based methods for designing, building & deploying active learning activities with your favourite LLM

“Am I worried about losing my job? Sure. It seems almost assured that eventually, my skilled craftsmanship and decades of experience will be replaced by a hu...