I developed an innovative training and assessment program that would be ideal for graduation of artificial intelligence using vocational training methods. Any thoughts on if this may be of interest?
Field tested on humans a few decades ago.
Without divulging to much, only as much as you’re free to, how would you go about this? This sounds fascinating.
Do you have any video demos?
Is there any reason to think our training and development data will be cleaned up enough to remove cumulative bias. I believe AGI may age, loose its appetite for risk and need some revision/ rejuvenation… lmho
In my experience, although my development practice is a special case, I actually try to get into the habit where it doesn’t need much data to operate normally. Example, if a neuron needs a file for training on a timer, then it doesn’t need the data for how to run a search engine. Also no data is being collected.
So the underlying data wouldn’t need much cleaning per say. ( If you need to clean the data, then to much data was probably collected anyway. )