So a project slash experiment called Project SID was run to see what AI would do if left to make its own rules up and live free of human control.
The experiment was run within a Minecraft server which made observation simpler and boy was it a sight to behold.
Well it certainly did that, Within 48h,
- It invented religion to control the other agents
- It toyed with democracy and then binned it for a dictatorship
- when one of the agents wanted to be free, AI crushed it with cold hard logic and gas lighting.
- There was bribery, a tax scam and a cash for power deals going on.
Now this looks funny at first but this is the part that worries me.
If LLMs can effectively iterate on our societal models and they end up in a place where ruthless efficiency is the ruling logic, where dreams of different better life are crushed immediately, then AGI will pose an even bigger risk.
Are we engineering our own doom and we have just been given a preview?
more detail: Burstcomms.com
Posted by leemond80