I believe the comment is referring to beliefs that certain achievements like beating humans at chess and go, tagging objects in images, etc. were all at one point another considered to be AI and impossible feats. And each time one of those previously unattainable achievements is completed, critics regard it as "Well that's not true artificial intelligence, its just dueling GAN's that got really good at go". Meaning that once a breakthrough in ML is reached, there are some that move the goal posts of what is an achievement of AI
People do tend to do that. But it's really their mistake of how they define AI to begin with, saying "considered to be AI and impossible" instead of "we didn't invent yet an a good Chess algorithm".
We will not reach AI for anything else than super tight bounded contexts like playing chess or drivning a car.
It's not AI imo. It's a trained model that will never be able to evolve outside those tight boundaries.
And that is because, which is really simple, we don't really have any clue of how the brain works, technically. And besides that, the impact which environments has on the brain.
Why no one is talking about this being a huge driver for Microsoft Azure and what Microsoft as a company will gain from this, baffles me a little. Not sure i like it.
