I watched this talk Ben Goertzel - Are We Building a Psychopathic Idiot Savant Global Brain? - YouTube and Ben is talking about the AGI network where AGIs exchange data & ideas very effectively. I was wondering before getting to AGI weather AI + Humans create preAGI. So AIs ask Humans for help and then they give the result / data to another AI which then can ask humans again. so in the beginning the AI is there to make the communication between humans more efficient and gather data from humans together. so its a human collaboration system
2 Likes
Yes, that is what can be called “Hybrid Intelligence”, here is my perspective on this for instance: Aigents for Global Brain
1 Like
That’s very similar to what I’m trying to find out and do if possible with a social learning group the wisdom of the crowd. Pair all the humans with an identical AI kernal and train each AI with the personal data set of that human. the core educational group will have data from all of it’s testing and opinion polls and another group project specifically designed to get personality data. An experiment could be done first without a group, using 16 AI programmed with a data set made from the 16 personality types from myers/briggs, and then have them do tasks, and maybe find which personalities work well together. I’m not really sure you can program a personality like that and have it stay that way without being paired with a human. Maybe they will just all act the same out of the gate. Maybe they will start out different and all learn to be like one of them. I don’t know but i think it could be a fun experiment to see how they interact with each other. If they even would be different. I wonder if i could just do a simple experiment with some virtual assistants that can work off line.
1 Like
We just have added Discourse forum integration taking this one as an example: Discourse social analytics and reputation graph analysis with Aigents | by Aigents with Anton Kolonin | SingularityNET
1 Like
Was wondering the same as @Mouyou . @akolonin: Good article and conceptual diagram. Do you know anything going on in this direction on an engineering, funding, or organizational level anywhere?
So you’re saying like an AI that when in trouble finding the answer to something, asks a human to try pointing out its mistake for it, and then sending that to another middle-man AI that organizes/analyzes that data for the first AI? Sounds really very interesting.