Why AI might fail its expectations?

When gauss as a boy was asked among his class to add from 1 to 100 by his teacher, he came up with the formula for adding terms of an arithmetic sequence. That's the kind of intelligence AI could never approach. He came up with a truly different solution. When La Place created his transform allowing extremely complex differential equations to be solved easily by transforming the whole problem from a domain to a completely different one, he just came up with a new efficient solution that AI can't just build with learning from feeding data into it.

Algorithms built to learn from data are fine, but they are way from creative.
And that's how humans really build their own experiences. It is not that straightforward.

At a 2017 conference of health IT professionals, IBM CEO Rometty told the crowd that AI “is real, it’s mainstream, it’s here, and it can change almost everything about health care,” and added that it could usher in a medical “golden age.” Guess what? Now only a few AI-based tools have been approved by regulators for use in real hospitals and doctors’ offices. They are just working as assistants to the doctors for routine tasks mostly to analyze images like X-rays.

AI could be a great technology. Yet it is just a tool to help humans with their higher hierarchy goals. You can read more here: How IBM Watson Overpromised and Underdelivered on AI Health Care.