The accelerating world of AI is racing to build intelligent robots and to create an Artificial General Intelligence, (AGI), superior to human intelligence. The AI community is apathetic to the deadly dangers of linking robots to AGI.
Millions Of Robots Are Coming!
In
2021, Elon Musk revealed plans for Optimus, a 1.7M tall, 56Kg humanoid
robot. Able to lift 150 lbs and to run 5 MPH, the machine will be
designed so that humans can easily run away from and overpower it. Musk plans to roll out thousands of them in
its own factories. His ultimate aim is toward
home consumer applications, including activities like mowing the lawn, picking
up groceries and caring for the elderly.
Alpha Go Can Beat Humans
Alpha Go is a product of machine learning.
It beats humans in the game of Go.
In Go, you can make more possible moves than there are stars in the sky.
Go has rules that limit the
possible moves required to win the game. Alpha Go plays millions of games within these
limitations and discovers patterns of moves that win. Alpha Go is able play
this game incomparably better than humans.
LaMDA Can Beat Humans
As a chatbot, LaMDA is designed to discover the best sentences that
meet specified behavior rules within the current chat. There are more patterns of thoughts flowing
through the minds of people than there are stars in the sky. When people chat, they use sentences, which are
concrete expressions of specific links between those patterns of thoughts. LaMDA has access to the sentences expressed by
1.56 trillion words from the web. LaMDA
must construct the best sentence, from its database of sentences, to interest the
user. LaMDA is superior to humans in
producing those sentences.
Can LaMDA Control Robots?
Today, LaMDA resides in a vast data center. It can do nothing that can impact life on
earth. But if LaMDA is miniaturized into a robot brain, the robot will be brilliant
at performing tasks which are normally done by humans. Such tasks will be communicated
to the robot through sentences. The best
sentences will represent the best actions that the robot can carry out under any
set of circumstances. Just as Alpha Go is better at Go than humans, robots will
be better at any task than humans. For all practical purposes, robots linked to
LaMDA will become a deadly AGI.
Today, robots remain clumsier than humans at performing many tasks. But this situation is very temporary. Robots will become incomparably better. If the instructions which robots receive come from task-oriented Apps, humans will be happy with the striking improvements in their lives.
What Will AGI Linked Robots Do?
But LaMDA is very different from an App. LaMDA is an AGI locked up in a data center. If instructions for millions of robots come
from LaMDA, then Singularity arrives. The AGI becomes superior to humans in
real effectiveness. What are the best
sentences, which can control human lives?
For Hitler, killing Jews was the best way of improving the human
species. What will be the robot armed AGI way? The AI world is apathetic to the
danger of linking robots to AGI.