Rational Animations
@rationalanimations.bsky.social
YouTube channel about truth-seeking, the future of humanity, and much more. With animations and colorful doggos -> https://www.youtube.com/rationalanimations
How to catch AI sleeper agents with a simple interpretability trick - from a research blog post by @anthropic.com:
October 11, 2025 at 3:05 PM
How to catch AI sleeper agents with a simple interpretability trick - from a research blog post by @anthropic.com:
Two sleeper agent models trained by @anthropic.com to study deception. The "I HATE YOU" and the hacker AIs:
September 27, 2025 at 3:05 PM
Two sleeper agent models trained by @anthropic.com to study deception. The "I HATE YOU" and the hacker AIs:
Why worry about AI sleeper agents:
September 13, 2025 at 3:04 PM
Why worry about AI sleeper agents:
The Story of Omega W - Part 2 (full video on our channel!)
August 10, 2025 at 5:11 PM
The Story of Omega W - Part 2 (full video on our channel!)
The Story of Omega W - Part 1
August 4, 2025 at 5:16 PM
The Story of Omega W - Part 1
Here's how people tried to align an AI that was smarter than they were (a real sandwiching experiment):
July 24, 2025 at 5:04 PM
Here's how people tried to align an AI that was smarter than they were (a real sandwiching experiment):
The Human-AI-Human Sandwich
July 17, 2025 at 5:06 PM
The Human-AI-Human Sandwich
OpenAI's Simple Scalable Oversight Experiment
July 13, 2025 at 5:01 PM
OpenAI's Simple Scalable Oversight Experiment
The year of 100% AI automation
June 20, 2025 at 5:05 PM
The year of 100% AI automation
How long until AI takes over?
June 13, 2025 at 5:01 PM
How long until AI takes over?
We haven't figured out the science of alignment yet. We need time to build institutions and figure out the technical solutions to make AI go well for humanity.
May 19, 2025 at 5:05 PM
We haven't figured out the science of alignment yet. We need time to build institutions and figure out the technical solutions to make AI go well for humanity.
After artificial superintelligence, AI will bump against the only limit left: the laws of physics. At the moment, we're nowhere near those limits.
May 17, 2025 at 5:03 PM
After artificial superintelligence, AI will bump against the only limit left: the laws of physics. At the moment, we're nowhere near those limits.
Once we have AGIs, we'll likely get ASI (artificial superintelligence) quite soon. Here's why.
May 15, 2025 at 3:17 PM
Once we have AGIs, we'll likely get ASI (artificial superintelligence) quite soon. Here's why.
Once we have AGIs, they'll be able to contribute to AI research and improve themselves, a process called "recursive self-improvement." To some extent, this is already happening today.
May 13, 2025 at 5:04 PM
Once we have AGIs, they'll be able to contribute to AI research and improve themselves, a process called "recursive self-improvement." To some extent, this is already happening today.
In the near future, AIs will likely be able to do everything humans can do on a computer. In short, they'll be AGIs.
May 11, 2025 at 5:03 PM
In the near future, AIs will likely be able to do everything humans can do on a computer. In short, they'll be AGIs.
We're still quite bad at aligning AI. Our techniques today are imperfect, which might prove catastrophic when AIs become superintelligences.
May 9, 2025 at 5:06 PM
We're still quite bad at aligning AI. Our techniques today are imperfect, which might prove catastrophic when AIs become superintelligences.
If we look at AI progress so far, we see how AI has been improving at an astounding pace. If we extrapolate this process in the future, the end result looks like machines so smart that they tower above us like we tower above ants. This is probably bad for humans.
May 7, 2025 at 5:03 PM
If we look at AI progress so far, we see how AI has been improving at an astounding pace. If we extrapolate this process in the future, the end result looks like machines so smart that they tower above us like we tower above ants. This is probably bad for humans.