Will we still be writing game engines in C++ [in ten years from now]?
No one can tell.
Is this situation going to last long, like in until-2022-long? Really, you shouldn't bet on it.
My belief is that one of the main technical reason why we still stick to C++ is memory management: with consoles being so limited memory-wise, you can't afford to leave the memory control to a virtual machine. So if next-gen consoles come with a massive amount of memory, C++ as we know it might very well disappear.
But it's even disputed if the next-gen is still a relevant concept. Maybe the living-room convergence people have been talking about for years will finally happen, and there won't be a market big enough to keep the AAA game industry as it stands now. Maybe cloud gaming will become the norm and games will be running on huge server farms that won't require such a fine-grained control of the machine.
Or, will John Carmack, write id Tech 7 in C++?
Still, no one can tell. Just look at the previous id Techs:
It's not even clear when will id Tech 6 come to life, and what it will do exactly, how could one predict which programming language an hypothetical id Tech 7 would be written with?
That said, as others answered, my feeling is that C++ is here to stay for a while, but I might be wrong. I suggest you read Scott Meyer's Effective C++, Third Edition, if you didn't already. The author explains how C++ contains already 3 different paradigms in one language (procedural, OOP, template-based): one could imagine a fourth paradigm becoming popular in the games industry in the near future, learning about it would be like learning a whole new language.
Now what you need to do is not to focus too much on the programming language. Of course becoming a C++ guru now gives you a good chance to get a job in the mainstream games industry. But what you have to master is software engineering, this will certainly stay viable for quite a long time.