The great danger of so-called artificial intelligence is not that computers will become sentient beings, but that decision-makers will treat them as if they are.
Machines are tools. They are a means to multiply human strength and to duplicate repetitive human tasks. They are highly useful. But they are not a substitute for human skill and judgment.
The use of automatic pilots in airplanes is a good example. An automatic pilot will make fewer errors than a human pilot, especially if airline management has pushed the human pilot to the point of exhaustion. But excessive use of automatic pilots means that the human pilot’s skills wither, and the human is less able to respond in an emergency that doesn’t fit the computer algorithm.
Another example is the use of the Internet and automatic answering machines for customer service. I don’t think anybody who has ever had to deal with one of these things thinks that they provide improved customer service. Their purpose is to create a barrier between the organization and the public in order to save money, but also in order to free the managers from the inconvenience of having to deal with actual human beings.
Machines don’t talk back. Not even self-directed machines talk back. Neither do they exercise judgment or think of ways to do the work better.
But from the standpoint of a bureaucrat whose goal is the seamless exercise of power, the latter consideration is unimportant. It is much more convenient to program machines than to deal with employees or deal with the public.