Predictions on the Arrival of Artificial General Intelligence

From Open Source Ecology
Jump to: navigation, search

Whoa, artificial intelligence can now (2020) recognize a cat, dog, car, and building.

There is still some time before artificial general intelligence [1] arrives. 2050 is a prediction. [2]

MJ's take on this: I predict that strong AI will arise only after humans learn to collaborate openly. Why? Because current learning by humans is extremely inefficient, in that the vast majority of state-of-art knowhow is proprietary, not shared, and used for militant rather than peacetime endeavors. This simply cannot and does not facilitate wide distribution of knowledge, learning, and prosperity - as the House of Wisdom attempted in medieval times. Only transparent, open knowhow - also recognized as open source - can allow for rapid learning. Unfortunately, humanity is deceiving itself today if it thinks that effective learning is taking place. This structural flaw must be addressed before higher learning - such as learning by machines - can occur.

This is because artificial intelligence is hard to do. It will require peak human intelligence, by definition, in order to attain. Such peak performance is incompatible with proprietary knowhow - which by definition - is scarce and hoarded knowhow accessible only to the few. Extreme human intelligence - as required to create artificial general intelligence - will require broad, interdisciplinary knowledge. Especially if it is to create good AI, as opposed to evil AI . This means that the agents creating artificial intelligence will need to be well versed not only in technology - but also in appropriate technology, history, psychology, humanities, economics - ie, build upon all of human knowledge - as opposed to mess the natural world order up even more. Some of the outcomes need to be social and environmental regeneration, and the distribution of prosperity far and wide - as opposed to the concentration of wealth and prosperity.