On Sunday, March 29, 2020, at 6:28 PM, Alan Grimes wrote: > the bottom line remains. We need artificial superintelligence. Good to see you are repeating my very words I said months back. Must have got in your head after some amount of context. Woho. 1up.
On Sunday, March 29, 2020, at 6:28 PM, Alan Grimes wrote: > We need artificial superintelligence. The definiition is this is an artificial intelligence capable of all modes of human thought and able to do it better as well as other capabilities not found in the human baseline. A system that is confined to one modality of operation fails to be even generally intelligent no matter how well it implements that modality. We need to make AGIs. They can easily become ASIs by being faster than humans, downloading tons of data, cloning brains, perfect recall, not pondering or sleeping etc lol, and a lot more. As for "all modes of thought", all you need is a Turing Tape 'sentence predictor' to manipulate it flexibly + more diverse data to get other "Modes". Images can describe all/any real world objects, text can too. As for non-human abilities, I already said some above, but we will find more to help its prediction ability and implementation ability. More diverse data helps improve intelligence :P too. On Sunday, March 29, 2020, at 6:28 PM, Alan Grimes wrote: > either a robotic platform or a rich VR simulation I've built my own text predictor from scratch. simply adding more text data improves prediction accuracy. And image predictors would be the same thing, generating stories, just better predictor with less data... All you need to make AGI is the small algorithm to generate stories with great accuracy, a desire to steer the story the way it wants, and a lot of data. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Td03911693a3a3f58-M674c728f5a54738d8de28781 Delivery options: https://agi.topicbox.com/groups/agi/subscription
