Right now, everyone is commenting that AI has trouble identifying images and using that failure as an all-encompassing example. But that’s something very specific, and something that only appears easy because of how hardwired it is in our own brains.

AI is mastering the smaller but more important skills, such as strategy. The best evidence for this is from the AlphaGo match, where the AI beat the world champion. And yet, people ignore the fact because it can’t recognize images.

Humans evolved to see, but imagine how difficult it is to teach a machine to recognize something, specifically something three dimensional. It’s near-impossible to code and relies entirely on machine learning algorithms. The fact that machines have gotten so good at recognizing things like faces should be worrying.

Now, let’s analyze what could completely wipe out humanity. Nukes probably won’t do it, as there will be people in bunkers hiding out. A solar flare might not do it, even if it destroys a lot of the asteroid. A meteorite could possibly do it, but we have good tech to turn it away. Even self-replicating nanobots would only destroy a single planet, because planets are surrounded by a vacuum.

AI could wipe out humanity. It could send self replicating probes to planets everywhere and it’s strategy would likely be better than anything a single human could come up with.

Elon Musk realizes that the best time to change this reality is NOW. AI is still developing and placing limits on it could possibly save millions of lives. It will be a while before any super intelligent machines emerge, but they’re coming and there isn’t any stopping them. Controlling them is Musk’s goal and with his initiative, humanity just got a little bit safer.