Zeroth Principle Thinking, Simulators, and Joscha Bach – Live and Learn #10
Welcome to this edition of Live and Learn. This time with essays on how to shape the future, my favorite talk from Joscha Bach, and an article describing what makes ChatGPT different from an AI security perspective. Enjoy.
✨ Quote ✨
Happiness is the result of being able to derive enjoyment from watching squirrels. It's not the result of understanding how the universe works.
– Joscha Bach
🖇️ Links 🖇️
Optimism Shapes Reality by Alexander Wang. If we are optimistic about the future we will work harder to get there. Furthermore, if we restrict the amount of time we have available to do something, we tend to increase our output. Being optimistic about how much time we need to achieve a goal then is a self-fulfilling prophecy that makes us work harder. Optimism shapes reality in that sense. I think there's a limit to this, but within reasonable amounts being optimistic and excited about the future, helps to get things done. Elon Musk is a master at this I think.
Simulators by Generative Ink. What is ChatGPT and where does it fit in with ideas from AI safety research? This essay argues that it is in a class of its own. Not an "oracle", "agent" or "genie" (terms with specific meanings in AI safety research), but instead a "simulator". Understanding why and how simulators can become existential risks to humanity – if they produce AGI – is a big, open question. And the type of "intelligence" that simulators like ChatGPT exhibit is quite foreign to us. ChatGPT is intelligent, but in such a different way than we are. It's like an alien brain we don't quite understand and this post tries to grapple with the problems this might cause.
Zeroth Principle Thinking by Bryan Johnson. There are ideas hiding in plain sight, which – if discovered – fundamentally change the way we see and interact with the world. Ideas like the concept of 0, computers, or general relativity. Finding those paradigm shifts is the most important activity humans can do. It's Zeroth Principle Thinking. It's discovering new rules of how reality works and then applying them. Bryan Johnson tries to live his life according to this idea. And he actually works on some awesome stuff like brain-machine interfaces. There's an interesting Lex Fridman interview with him, which I highly recommend listening to as well.
The Ghost in the Machine by Joscha Bach. Probably the most informationally dense talk on YouTube. In one hour, he answers deep questions such as: What is knowledge? What is intelligence? And what is consciousness? The main idea built up from that is computational functionalism: Our brains are hosted by a physical world. They gather sensory data, which is used to produce a model of the universe. Our brains also put a self (us) into this glued-together "dream world". This self can act, and those actions are translated back into the real physical world via our bodies. Actions lead to new sensory data and our brains try to keep the error between the expected state and the sensory data as low as possible while also trying to seek out states beneficial to our survival. These ideas are very similar to how consciousness is described in books like "Being You" by Anil Seth.
🌌 Midjourney 🌌
🎶 Song 🎶
Desafinado by Stan Getz
I hope you found this newsletter useful, beautiful, or even both!
Have ideas for improving it? Please let me know.