🎙️ Join the 80,000 Hours Podcast team to help shape the most important conversation of our time. We’re looking for a new host + chief of staff to make great content on AI and the intelligence explosion. Register interest by Apr 6: 80000hours.org/2025/03/podcas…

x.com/i/article/1903…

1/ People are saying GPT-4.5 is disappointing, but it looks on trend to me (fight me). On difficult scientific reasoning questions (GPQA): GPT-3.5: 28% GPT-4.0: 42% GPT-4.5: 69%

AGI by 2027? I spent weeks writing this new in-depth primer on the best arguments for and against. Starting with the case for...🧵

Cutest illustration of AI progress: "otter on a plane using wifi" 2022 vs. 2024

What can experts tell us about when AGI will arrive? Maybe not much. Except that it's coming sooner than before. I did a review of the 5 most relevant expert groups and what we can learn from them.. 🧵

AGI timelines have plummeted from 50 years to 5:

o3 is on the faster time horizon trend that started in 2024, suggesting AI is moving ~2x faster than before, and possibility accelerating.

A new legal letter aimed at OpenAI lays out in stark terms the money and power grab OpenAI is trying to trick its board members into accepting — what one analyst calls "the theft of the millennium." The simple facts of the case are both devastating and darkly hilarious. I'll

Why to quit your job and work on risks from AI (the short version) 🧵

How far can reasoning models scale? "Available evidence suggests that rapid growth in reasoning training can continue for a year or so." epochai.substack.com/p/how-far-can-…

This is very funny, and good advertising for shrimpwelfareproject.org and effective altruism. youtube.com/watch?v=VNbIKt…

Scholarship to develop AI tools for navigating high-stakes, complex, urgent situations: "...fellowship on AI for human reasoning to help talented researchers and builders start working on AI tools for coordination and epistemics" 1/

Funding opportunity to work on understanding and managing risks from internal deployment of frontier AI models (an often ignored issue). 1/

I'm interviewing Toby Ord, author of The Precipice, on: • AI governance implications of throwing more compute at inference enabling models to act more intelligent • Practical implications of reinforcement learning (and reward hacking) making a comeback • His article

Next week I'll be interviewing Hugh White — Professor of Strategic Studies, and author of 'Hard New World: Our Post-American Future' — about: 1. The radical changes Trump is bringing to geopolitics 2. How erstwhile US allies will react to the withdrawal of American friendship 3.

We’ve just published a new AGI resource hub, designed to help you figure out: • what’s going on with AI • what the biggest societal-scale risks might be • how you can help. Find out how we might be able to help you get started with working on this, or get up to speed with the

US allies have long lent on its military for protection. Trump is hell-bent on shattering that dependency — and abandoning US hegemony. Professor Hugh White has predicted this retreat for 30 years and despite calling the US-led order a 'golden age', he thinks it's a good thing.