You really first should decide your target platform that you are most interested in, and if your primary goal to work on other people's games at a game development company, or your own games for side projects hoping to start your own company one day.
On platform, I recommend mobile for a number of reasons. It is a rapidly growing market with much less saturation then PC, and gamers are more accepting of hobbyist games than on other platforms. Also, job-wise the number of mobile game dev jobs is skyrocketing while there is less growth for jobs in other platforms (though they are at a much higher job base). You just have a lot less competition on mobile right now, and that makes it easier to get hired and also move up the ranks as you are competing with people with < 5 years in mobile versus often 20 years on PC/console.
If you go to mobile route, I highly recommend CoronaLabs / Lua to learn first. It is easy to learn and get a game going much quicker than almost any other environment.
Many new game devs believe getting into something with a higher learning curve, but that is used more often by some "hardcore" professionals, is a way to shortcut the process. I strongly disagree! The biggest problem by far for new game programmers is that they fail to ever produce any playable game. Your resume and skills are way more impressive with a successful iOS game that is actually selling (even if only $1000/year) written in CoronaLabs, than someone who has strong C++ game dev skills on paper but has never produced a public game.
This isn't to say you should never learn C++, I'm just recommending Corona for your first couple of games-- then at that point you can either keep going with it (as you may find success with that route), or at that point start learning C++. But it is really important to build a strong foundation of game dev skills first, and not trying to make life difficult by trying to learn game skills in a language/sdk tool set (C++) that requires a far steeper learning curve.