Ben talks about neural symbolic architecture for cognitive robotics. Ben shows demonstrations of physicalized AGI systems using neural symbolic architecture using the Nao robot platform. Doc Goertzel shows video of an AGI which recognizes the emotions of a human, and proceeds to show video of a Second Life implementation of AGI. Ben goes on to list a proven record of AGI (Biomind) assisting in medical research. For example Biomind was used aging networks from gene expression data from Genescient’s Methuselah flies.
Hameroff takes the stage and remarks on the historical failures of artificial intelligence research, citing Newell and Simon (dating back to the 40s.) Not because it’s impossible but because the researchers are not modeling the correct structures. He goes on to describe microtubules, quantum effects in the brain, gamma-states indicative of consciousness (*BING*) states in the brain. He glosses over the Quantum Consciousness theory he has developed with Roger Penrose and encourages AI researchers to model cilia.
The SIAI’s (Singularity Institute’s) Anna Soloman just finished the opening talk about intelligance and the Institute’s vision for a controlled intelligence explosion, as opposed to an uncontrolled intelligence explosion that would destroy us and everything we value.
Anna discussed the gradual technological progression for intelligence that will eventuall make humans obsolete, and described a number of avenues for incremental research progress, so that we can eventually learn how to build an intelligence that we understand, and that will create a world we value.
In order to emulate a brain we’re going to need greater scanning capability:
• need enough resolution – rough consensus 5x5x50 nm resolution scanning
• need enough information
• need enough volume
“The step from mouse to man is 20 years in terms of brain emulation. First scan or simulate then computer power gradual emergence of emulation.”
Randal is impressing upon his audience the importance of shedding the fleshbag.
Theodore Berger building hippocampus replacement.
“In-vivo techniques, neural recording, neural interfacing:
• Scary/risky procedures
• chronic implantation
• power supply
• scale and bandwidth ..”
Different technologies could provide less-invasive (?) modalities: qdots.
Multiscale scanning requirements, need to record neuronal activity at different levels of activity (voxel, group, spike, analog, spatial, molecular)
Deep Machine Learning
• biologically-inspired computational intelligence approach
• adequate tech is here
• billions of transistors on a single chip
• power requirements are low
Ben Goertzel asks Arel what his funding requirements are … Arel says he can build you an AGI system for 2-14 million dollars.