Feel the ASI
We are approaching a pivotal moment in human history,
and we need your help to navigate it responsibly.
A growing consensus of notable AI researchers, tech leaders, and public officials believe that highly capable AI systems, possibly even superintelligence, could arrive within the next 5-10 years.
It sounds like science fiction, but it's a possibility we all need to take seriously. AI systems are approaching the ability to improve themselves without human help. Once that happens, progress could accelerate beyond our ability to predict or control.
What Leading Voices Are Saying
Scientists, tech leaders, and elected officials across the political spectrum are raising the alarm.
Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.
average probability that advanced AI leads to outcomes extremely bad for humanity.
Image credits
Gates — Kuhlmann/MSC, CC BY 3.0 · Hinton — Xuthoria, CC BY-SA 4.0 · Hassabis — John Sears, CC BY-SA 4.0 · Altman — Steve Jurvetson, CC BY 2.0 · Amodei — TechCrunch, CC BY 2.0 · Pichai — Lukasz Kobus, CC BY 4.0 · Musk — Duncan.Hull, CC BY-SA 4.0 · Hawking — Jim Campbell, CC BY 3.0 · Obama, Vance, Buttigieg, Sanders — official portraits, public domain.
Essential Resources
International AI Safety Report
The most authoritative global scientific assessment of AI capabilities and risks. Includes an accessible 3-page executive summary for non-experts.
The AI Revolution: The Road to Superintelligence
A classic 2015 exploration of AI, exponential growth, and why superintelligence matters. Still highly relevant and accessible.
The Problem
A clear explanation of the core challenges in AI alignment and why they matter.
AI 2027
A well-researched fictional scenario exploring plausible future outcomes in the race to AI supremacy. See also the AI Futures Blog for ongoing updates, forecast tracking, and policy analysis from the same team.
Situational Awareness: The Decade Ahead
A comprehensive analysis of the path from GPT-4 to AGI to superintelligence, including national security implications.
If Anyone Builds It, Everyone Dies
A New York Times bestseller making the case that building superhuman AI with current techniques poses an existential threat to humanity, and why the situation is still preventable.
Should We Pause AI? Here's the Debate.
If superintelligent AI could cause human extinction, why don't we simply stop building it? A clear, animated overview of the main arguments, practical difficulties, and proposed responses.
AI Is A Massive Problem. Here's Why.
A thorough walkthrough of how modern AI actually works, why it's advancing so fast, and what the risks are. Well-produced and accessible even if you have no technical background.
Taking AI Doom Seriously For 62 Minutes
A thorough, well-produced deep dive into AI existential risk.
AI Explained
Stay updated on major AI developments with balanced, sensible analysis from the creator of SimpleBench.
AI in Context
Thoughtful analysis and commentary on AI developments, exploring both the technical and societal implications.
CivAI: Explore AI Capabilities
Interactive demos that let you experience what AI can actually do, from creating deepfakes in seconds to generating personalized phishing attacks, so you can understand the capabilities and risks firsthand.
BlueDot Impact Courses
Free courses on AI safety and governance designed to sharpen your thinking on what actually matters. No technical background required for the governance track.
Epoch AI
The gold standard for empirical AI trend data. Interactive dashboards tracking compute growth, model capabilities, and hardware trends. Hard numbers instead of hype or speculation.
AISafety.com
A curated portal for the AI safety ecosystem: field map of organizations, job boards, study courses, community links, events calendar, and funding directories. The best starting point if you want to explore what's out there.
One Useful Thing
A Wharton professor's hands-on exploration of what AI can actually do today. Concrete demonstrations and clear-eyed analysis that help you grasp the real capabilities and limitations of current systems.
AI Futures Project Blog
Ongoing analysis and forecasting updates from the team behind AI 2027. Follow along as they track how reality compares to their scenario, grade their predictions, and explore what comes next.
Target Curve
A blog exploring why automating intelligence is fundamentally different from previous technological revolutions, and what it means for society. Covers the evidence, addresses common objections, and focuses on what we can actually do about it.
Planned Obsolescence
Practical writing on AI preparedness from a leading safety researcher, focused on concrete milestones, timelines, and what we should actually expect.
Take Action
You now know what's at stake. Here's how to make a difference.
Make a Microcommitment
One small action every week — delivered to you. Share an article, sign a petition, email a representative. Never wonder what to do next. Just 5 minutes a week.
Add Your Name
Sign open letters calling for international AI safety standards. Takes seconds, and every signature strengthens the case for regulation.
Contact Your Representatives
Tell your elected officials you want real AI oversight. ControlAI provides pre-written templates. Just fill in your details and send.
Join a Community
Connect with others who take this seriously. Organize local events, coordinate advocacy campaigns, or contribute to grassroots efforts pushing for a pause on frontier AI development.
Learn AI Safety
Take free courses on the technical and governance challenges of advanced AI. BlueDot Impact runs cohort-based programs that have trained thousands of people now working in AI safety.
Work on AI Safety
If you want to dedicate your career to AI safety, 80,000 Hours offers free one-on-one advising to help you find high-impact roles in AI safety policy, research, and governance.
Never doubt that a small group of thoughtful, committed citizens can change the world; indeed, it's the only thing that ever has.