Knowledge download is one of two components of learning process (another component is Experiment).
Natural language text can be read by text
parser. Extracted information can be put into the
main memory.
Ability to read gives strong AI access to such information sources as
electronic books, web pages, emails, chats, etc.
Software developers can create additional devices for Strong
AI. Such devices can have its own data storage. There
could be special algorithms which would collect and put external information
into these special data storages.
Example:
special strong AI device may allow fast search on the web. Google’s index
may be imported into such web-search device.
Main memory can be updated with specially prepared chunks of data. For example, such chunks of data can be copied from one instance of strong AI to another. It may be useful to quickly copy knowledge from strong AI #1 that spent time in experimentations about topic XYZ into strong AI #2 which didn’t have XYZ experience yet.
Knowledge download is very fast and powerful, but alone it is useless for learning process. After knowledge download strong AI needs to follow-up with some experiments. The reason for that is that in order to become useful, knowledge must be connected to goals and other hardcoded units.