Minsky is right. Sure, today's AI solutions uncover credit card fraud: but they are narrow specialists. Only capable of attacking tightly defined problems.
The big goal, the creation of artificial intelligence (and dealing with the big picture and the overall question of what intelligence actually is) has receded into the background, replaced instead by a focus on details and fragments of the world.
Logically, there's money for this from the business sector - in times when government funding for research without a specific purpose is declining further and only purpose-driven private funding (or from the business sector) is available, researchers eventually have to deliver something as a result that can be put to use.
And that's why real AI research is falling by the wayside. We now understand in much greater detail how computers can solve partial areas. But do we know more about how intelligence actually works? How we can replicate such a thing with machines? No. That part has receded into the distant future.
At heise online news there's the original article.