CS 1501: AGI Lecture 7 | Alternate Paths to AGI

“The winds, the sea, and the moving tides are what they are. If there is wonder and beauty and majesty in them, science will discover these qualities... if there is poetry in my book about the sea, it is not because I deliberately put it there, but because no one could write truthfully about the sea and leave out the poetry.” 

― Rachel Carson

Course Recap:

  1. Besides cognitive architectures, there are some other ways to theoretically achieve an AGI. We talked about whole brain emulation and Seed AIs.

  2. Whole brain emulation involves uploading a human mind to a powerful computer by means of scanning, translation, and simulation.

  3. Advantages of whole brain emulation involve AGI warnings and the fact that we do not need any major theoretical breakthroughs, just technology.

  4. Turing had the that instead of creating an adult mind, we should create a child mind and let it learn things.

  5. This led to the concept of a Seed AI, but is different in the sense that a Seed AI is a small initial program that can improve its own code.

  6. Opens the door to two fundamental concepts:

  7. Recursive self improvement: where a program gets exponentially smarter by modifying its source code/architecture iteratively

  8. Indirect Reach: That AGIs have the capability to impact way more than initially thought due to recursive improvement.

View Full Lecture PDF Below:

Week 7_ Alternate Paths to AGI-2
Download PDF • 2.02MB

2 views0 comments