Remember DALL-E is good at predict what is beside 4 different shaped pikachu teapots? Another different pikachu teapot.
I believe this is as simple as being shown a movie example of folding towel A, then shown a towel B in real life - its brain will predict the recalled movie but this time folding the similar towel a similar way, this will trigger needed motors. To get linked motors it can simply thrash arms around for some minutes, so collect some sense2motor "pairs" - not that that is CLIP related "pairs", just simple links. You Don't need a motor cortex, just a good multi-modal predictor of the thoughts man. Motors are just a link to move them for REAL. The Cerebellum smooths actions so they aren't choppy, so you can do 4 changes to motor rotation direction and speed per sec and look like you do 64 actions per sec. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T7fc016f9b4c5d969-Meb60860ca125c2d69a4cd0ac Delivery options: https://agi.topicbox.com/groups/agi/subscription
