On: Surviving the Cosmos

Episode: 22

Date: December 2015

Key Subjects:

• Professor in Physics, author of “The Fabric of Reality”, “The Beginning of Inifinty”.
• Knowledge:
  • A kind of information.
  • It says something that is true and useful about the world.
  • The generic thing for converting some matter into some other matter. (see also “Why Information Grows“)
  • Not necessarily tied to any specific individual (generalized).
  • Can exist independently (until falsified).
• Science:
  • Not: empiricism, every theory needs to be testable in the real world (falsified or verified).
  • But: which theory provides the best explanation.
  • Faith in science relies on faith in the strength of the error correction mechanism – to what degree are experts challenged.
• Power of explanation:
  • Anything which isn’t precluded by the laws of nature is achievable given the right knowledge.
  • Because if something were not achievable, given complete knowledge, then that itself would be a regularity in nature which could be explained in terms of the laws of nature.
  • So, either something is precluded by the laws of nature or it is achievable with knowledge.
  • Restrictions in practice: the actual laws of nature and limits in a person’s knowledge.
• Reach of explanation:
  • Parochialism: knowledge expansion is limited – a human/chicken/plant evolved to occupy a specific cognitive niche (and not more).
  • Incorrect: anything that can be explained – either in practice or in principle – can be explained (by us).
    • Universality of computation: given the right program, we can perform any kind of transformation on information, including creating knowledge (if we know how).
    • Only limitations are memory capacity and speed [which gets you back to parochialism?].
• The importance of error correcting systems in cultures:
  • Knowledge is conjectural and subject to improvement.
  • Protecting the means of improving knowledge is more important than any particular piece of knowledge.
  • In closed cultures, there is no room for institutionalized criticism.
  • Institutions that, in contrast, suppress the growth of (moral) knowledge are immoral.
• Morality:
  • Morality is about the problem of what to do next.
    • A coherent set of ideas.
  • Morality, like knowledge, evolves by error correction.
    • Adapts to cultural environment.
• Benign view on AI:
  • Badly summarized: its morality and behavior would at least initially be similar to ours.
  • If anything bad may happen related to the invention of an AI, they are exactly the same things that we have to watch out for even if AI doesn’t happen.
  • The “normal” error correcting mechanisms of human culture would apply.
  • Good and bad outcomes are both plausible.

Key Takeaways:

• Knowledge: something that is true and useful about the world.
• Science: which theory provides the best explanation.
• Faith in science relies on faith in the strength of the error correction mechanism – to what degree are experts challenged.
• Morality is about the problem of what to do next; like science, it evolves by error correction.

Worth Listening:

8/10