- Here's the deal, no longer do just experts or star gazers make predictions. Predictions can be made by machines and massively so, using neural networks and deep learning algorithms. I'm not going to tell you exactly what they are or the math behind them but it's fun to explore the prediction capabilities that machines have today. Let's start with the really simple stuff. You know this, everybody does this, start typing something into the Google search bar. It'll automatically give you a dropdown of what it predicts you want to search.
Just stop to think about it for a second. It's taking into account your geographical location, previous searches, live world events, bookmarks, visited sites, and so much other information. And from that, it's making predictions about what you may be searching. That's pretty incredible. Let me give you another example. I was walking through a museum in Boston and I saw an exhibit that fascinated me. There are machines that can take DNA off the most disposable objects. For example, cigarette butts and reconstruct the face of the person that smoked that cigarette.
Legitimately mind-blowing. I don't know the exact accuracy of these predictions so let's take that with a grain of salt. Another thing you know and probably hate to love, that little segment on your Amazon account landing page that says, items you might like, or recommendations for you, it's all being done by machines. These powerhouse machines that are constantly reading you, trying to market to you, and then even learning from their mistakes from when they're wrong. I heard a joke the other day.
A man asks a machine, what's 11 times 11? The machine replies, it's 58. The man says, ha, you're so wrong, it's 121. The machine replies, it's 121. I never said it was a good joke but you get the point. Machines will always keep learning if programmed to learn, and they'll make better and better predictions throughout their life span. Another outstanding issue is that neural networks and machine learning function inside a black box. Imagine that you feed the black box a problem, you can't see what's going on, how are you able to trust that output? Where do we draw the line? When do we start trusting machines? Is there a replacement for human intelligence and experience? But we suck at making predictions.
Are machines better than us? I don't know, there's so much to ponder, so much to overwhelm you, so many ethical implications. Next, I'm going to interview Adam Geitgey. Adam is an AI consultant and I'm sure he has a lot of opinions about how predictions are made by humans and machines and how these predictions affect the general public.