Thread

1. amelius · 1791061907 · focus · HN ↗

   How to know if a machine can gain consciousness:

   1. Write a simulation environment like conway's game of life, perhaps a little more complicated. But leave out any human bias.

   2. Run it until life starts to form.

   3. Wait until the creatures start speaking about the concept of consciousness (of course in their own language, which you'll have to decipher first).

   The way Amodei did it is with human bias, so it's not going to work.

   1. pmontra · 1791084601 · focus · HN ↗

      That 1-2-3 framework can't be applied to LLMs because they are full of human bias by definition, as they are trained on human language. We will have to find another way.

      Alphago is an example of an AI that has no human bias, because it learned go with self play without seeing any human game.

Unofficial Hacker News client; not affiliated with Y Combinator.