When it comes to receiving information on topics like; Artificial Intelligent, Machine Learning and Data Science, there is no better way of learning than reading. The reason is simple — it’s great in terms of time and quality of absorbing information. Reading is also subjective and up for interpretation, which makes for interesting conversation and debates. However, when it comes to video and images, we’re confident that what we are looking at is factual. It’s a fact because you’ve seen it, right?
Well, in today’s world you may want to reconsider the lens you’re looking through. I would start by looking at the person attached to this link: thispersondoesnotexist.com then reload the page a couple of times and look again.
What did you think? They look like nice people, right?
I’m sure at this point of reading scepticism is high and it should be. The faces you looked at are not real people, and never will be. They’re developed by Software Engineer Philip Wang, who built the site with algorithms called generative adversarial networks (GANs). Every time you refresh the page, the network generates a completely new facial image.
I found myself mesmerised by this. I think we all naturally, and very quickly, spin up this background story of someone when we look at someone, especially looking into the eyes of a person for the first time. After all, “the eyes are window to the soul”.
It’s haunting looking into the eyes of these “people” and knowing they have no story, they have no past, no present, no future. They’re just data presented in a way that looks human.
Without getting too metaphysical, if nothing else I wanted to draw your attention to something on the horizon. Recently, Google Assistant has showcased booking appointments to real people over the phone.
Now, as Philip Wang’s algorithm has shown us, it’s not an unimaginable leap to see a future where we don’t even realise we’re talking face-to-face with a machine. Scary or incredible?
If nothing else it’s interesting and worth thinking about next time your eyes scan across an image.
