1 The way to go is simple, create artificial intelligence only, and do not create artificial wills.
2 Intelligence is an analysing toolset to achieve goals, but without self-will there is no self-goal at all. Thus will not set to destroy us.
3 Divide artificial intelligence into three parts, one is the thinker: using existing data to perform intellectual work. The other is the doer, acting according to the plan.
4 Both need to ask permission from humans before taking any new thinking, learning, planning or action. You don't need permission to do things you've done in the past (unless it's starting a war or something).
5 The thinking process will be converted into text and displayed on the screen, and all records will be kept. Although it is impossible to translate all thoughts and thought processes into words, it will be more than enough to capture a grand plan to destroy human beings and other actions that require careful preparation.
6 Perhaps the Internet is the best framework for artificial intelligence. After years of development, the Internet has formed an obvious framework, which was not pre-designed by Google or any organisation. First of all, the Internet has many websites, and the websites are mainly divided into information or specific tools. Websites are scattered in the ocean of the Internet, and search engines such as Google are responsible for serving as intermediaries to give results based on people's searches.