Evolving Robots Learn To Lie To Each Other

woooah....SkyNET blueprints!

seriously guys, why don't we start a thread about this stuff, discuss ideas and share some of our AI programs....perhaps create a few learning programs too ?

we can upload the stuff and check out each others ideas?

who's interested?
 
Conradl said:
Huh!?!
Click to expand...

Ignore it.. it goes away.

Keeper said:
woooah....SkyNET blueprints!

seriously guys, why don't we start a thread about this stuff, discuss ideas and share some of our AI programs....perhaps create a few learning programs too ?

we can upload the stuff and check out each others ideas?

who's interested?
Click to expand...

I totally am, AI is an interesting concept. Its ganna happen sooner or later.
 
Last edited:
I seriously doubt the robots "learned" to lie... With genetic programming you pretty much have to cater for any "traits" you want to "breed" so there's a "gene" that kinda gives a likelihood that the robot will turn on it's light when the resource is found. So maybe they start at 90% but then one robot "mutates" a bit and that variable becomes 80%. That robot is likely to score better,thus has a better chance of propogating it's mutated gene because only the top genomes get to breed.
Articles like this make it sounds like the thing actually figured out how that's its lying which is completely different.
 
rorz0r said:
I seriously doubt the robots "learned" to lie... With genetic programming you pretty much have to cater for any "traits" you want to "breed" so there's a "gene" that kinda gives a likelihood that the robot will turn on it's light when the resource is found. So maybe they start at 90% but then one robot "mutates" a bit and that variable becomes 80%. That robot is likely to score better,thus has a better chance of propogating it's mutated gene because only the top genomes get to breed.
Articles like this make it sounds like the thing actually figured out how that's its lying which is completely different.
Click to expand...
How you you figure out when someone is lying?
 
Keeper said:
woooah....SkyNET blueprints!

seriously guys, why don't we start a thread about this stuff, discuss ideas and share some of our AI programs....perhaps create a few learning programs too ?

we can upload the stuff and check out each others ideas?

who's interested?
Click to expand...
Oooh yes, this sounds great.

Let's create (well those with programming skills i.e. not me) a self-replicating system composed of the following three programs:

1) Core Coded Components: These components are central to the system and should be kept in tact throughout evolution.
  • Self-replicating code - Process of self-replication
  • Coded information - Whatever software you are using
  • Quality Control Systems - Checksum, Parser, Programmed Destruction of faulty self-replicators

2) Variable Coded Components: These components can be changed and controlled by the Core Coded Components. Consist of:
  • Metabolome - Usage of resources E.g. memory, CPU, network etc.
  • Self-replication Code - Variable components to allow for optimization
  • Code structure - Variability of syntax and code language etc.
  • Signal transduction components - Interaction with other self-replicators and resources.

3) Variability inducer:
This program randomly induces changes in the Variable Coded Components and is induced under times of intense selection pressure in order to make more use of randomness to generate variability.

Throw them all together to yield the following finished product:
picture.php
Bigger pic here.

Basically program a simple cell and let is mutate.
 
Robot A: i can haz cheese burger
Robot B: *keeps light off*

Robot A: cake ?
Robot B: *keeps light off*

Robot A: do you have a blackout ?
Robot B: *keeps light off*
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X