1. Notice that the story [of technical progress accelerating indefinitely] is not testable; we just have to wait around and see. If the predicted year of true AI's coming is false, too, another one can be forecast, a few decades into the future. AI in this sense is unfalsifiable and thus--according to the accepted rules of the scientific method--unscientific.
The Myth of Artificial Intelligence: Why Computers Can't Think the Way We Do
2. This cuts the myth at an awkward angle: it is because the [artificial intelligence] systems are idiots, but still find their way into business, consumer, and government application, that human-value questions are now infecting what were once purely scientific values.
The Myth of Artificial Intelligence: Why Computers Can't Think the Way We Do
