"Artificial" intelligence in humanoid form will prevail where interaction with humans is the primary focus.

Most people prefer to interact with what is familiar to them.

Examples of this include communication interfaces with AI but also entertainment such as cinema, TV, prostitution etc.

AI will present itself in other - more efficient - forms when it comes to processing factual matters.

The crucial question is not what AI can achieve.

But rather what humans could still achieve that could not be sufficiently achieved - i.e. simulated - by AI.

The key word is "sufficiently" - which is not the same as "better".

And whether this is enough to justify human precedence and its costs.

Humans must rethink their goals - their justification for existence.

If they do not do this or if they cannot find an answer to this, the new intelligence will do it for them. That is how evolution works. And that is a good thing.

AI is based on the statistical use of masses of data and on more or less validated algorithms. It does not reason, nor does it understand. It may perhaps offer leads, but never understand. And it cannot claim to rival the robustness of millions of years of biological evolution or decades of patient scientific and humanistic learning. Let us beware of computers and software of all kinds, chaotic deterministic systems, subject to daily bugs. And above all, it relies on robots devoid of passions and therefore of emotional intelligence, an essential driver of all creativity and admittedly a Pandora's box. Human thought must remain the yardstick of human life, for better or for worse. At worst, AI, like social networks, is a machine for dumbing down. Did you say intelligence? "Faculty of knowing, of understanding" (Robert).