When deep learning turns into deep loving

Being in love with a computer or an application. It is already widely discussed but this phenomenon is closer than you’d think.

Originally posted on September 30, 2016
By Paul Schoenmakers

Her

If you have seen the movie ‘Her‘, you would know what future possibly lies ahead. If not, go watch it. Without spoiling too much, the movie is about a guy who falls in love with a highly developed version of Siri, the virtual assistant for the iPhone. Without ever seeing an appearance of her, simply because there is none, the poor guy gradually grows into a relationship with this operating system until there is no way back.

Now think about online dating. People often only talk to each other on Whatsapp or similar applications for the first couple of weeks. From personal experience, I could say that — even when you haven’t seen her in person — it is possible to feel some kind of connection. Of course, you have seen some vague pictures of her from flattering angles but you really start to like her for her personality. Obviously, there is a difference between a person that is typing and an application that is giving pre-programmed answers. However, the crucial factor is that these applications are getting more advanced. This is where machine learning and deep learning come in.

Paula can see

In 1959, the term machine learning was defined as a “Field of study that gives computers the ability to learn without being explicitly programmed“. Basically, this means that a computer could recognize patterns and make data-driven predictions or decisions. A virtual assistant, let’s call her Paula, could behave more human-like with the help of machine learning. For example, when Paula is released and you ask her what she thinks of the American presidential candidates Clinton and Trump. Probably the answer is not pre-programmed so she doesn’t know what to say. However, when more and more people discuss these two candidates, Paula could analyze the public opinion. The next time you ask her, she has learned to response appropriately. Of course, it is not really Paula’s opinion but isn’t almost all our opinions also influenced by external factors?

One step further is when she recognizes your facial expression and adjusts her ‘opinion’ to your mood. Deep learning, a branch of machine learning, is making this possible. In deep learning, images can be analyzed and this will add a dimension to the virtual assistant. Now Paula can see! A famous example is where the deep learning algorithm sees an image of a guy in a black shirt playing guitar. It has never seen this particular image but it has seen a man before, it has seen the color black before and it has seen a guitar before. The output of the algorithm was: ‘man in black shirt is playing guitar‘. At first is may not seem like a big deal, but if you think about it, this is similar to how people learn!

What is love?

These are just some of the possibilities of machine and deep learning in virtual assistants but is loving ‘her’ not a little far-fetched? No, I don’t think so. Love is just a chemical reaction in our brain and people have been falling in love with the weirdest objects. I have heard of people falling in love with pillows, animals, mannequins and even a piece of the Berlin wall (yes, really). Suddenly, falling in love with a human-like responding computer doesn’t seem that weird anymore, does it? Of course, I can imagine you are thinking: “something that talks back is nice but I really need an appearance to fall in love with“. Yes, I agree, but human-like robots are also emerging.

Get in touch: whatsup@oneup.company