Humans are story-tellers, and for hundreds of years we've been trading ideas and teaching generations through the medium of narrative. Story-telling techniques are powerful tools, and consuming fiction for entertainment is a past-time you'd be hard-pressed to erradicate.
But what happens when fiction is presented as fact?
Generally, given enough life experience and a critical approch to information, most people are capable of discerning truth from fabrication. If I tell you there's an elephant balancing atop Big Ben you'd probably laugh - the idea is so ridiculous it's easily filtered out from fact.
But what if I told you that a silverback gorilla had escaped from Chester Zoo? With just enough truth certain statements trigger the sliding scale of rationality to shift. Chester Zoo certainly has a silverback. Animals have been know to escape before, and silverbacks are intelligent, powerful creatures.
What if I showed you a video of a silverback climbing over the zoo's fence?
With the spoken word and printed page (or backlit pixel these days), it's easier to dismiss claims. But with images, claims begin to take on new life.
So powerful is the medium that it's been known to spawn decades of folklore and debate. The idea of a Loch Ness Monster gained serious traction after a photograph taken in 1934 began to circulate - after all, seeing is believing.
Quite a lot has happened since that famous photo was taken. We've been used to airbrushed models and advanced CGI in our films for years now (to the point that the latter has become old hat), but there now exists technology that doesn't require years of training and hours and hours of skilled time for something fantastical.
This video is the output of this open-source project and was created by training the software to map horses to zebra, and then giving it a video of a horse. This technique mimics the human brain's ability to form complex models of reality in our heads, and then apply those models to the real world.
As an example, take a look at this photo:
Having never seen the image before you're still probably able to discern:
- That there is a horse,
- There is also a human,
- The human is caring for the horse,
- There is a tool involved in this action.
The above analysis happens without conscious thought thanks to the advanced networks of recognition we have stored in our heads. The software applied to the video above uses the same principals to build an understanding of the world.
The video above isn't perfect, but it's pretty interesting. Given that this technology is in its infancy, in a short time it'll be hard to tell the difference between fabrication and reality.
Why Crazy Realistic Zebra Horses Matter
Sure, paining a horse in real-time to look like a zebra is cool, but how does it impact me?
The above video demonstrates the sheer power of a similar technology. In a short period of time, it's possible to build out very realistic video and audio without a requirement for techniques such as 3D face mapping and advanced voice analysis.
And while the tale of a loch's mysterious inhabitant is certainly interesting, the rise of questionable media (see everybody's faviourite topic: fake news) hints at a dangerous undercurrent that might cause a dramatic shift in content production in the near future.
No longer do we seem to care whether or not something is factually accurate. The value of content is now measured by the number of likes, clicks and re-posts it garners.
At least initially the technology will be used to amuse - if I could make any celebrity say something outlandish and funny, I'd probably give it a whirl.
But what happens when a bad actor picks up the tool, subtely tweaking content and dropping in an inflamatory phrase? Combined with social media's perchant for viral content, something entertaining could spread extremely quickly.
What if I sent you a recording of your best friend disowning you, and used it to manipulate you?
Soon, the loch will be overflowing with monsters.
Only they'll be in Ultra HD, and be indistinguisable from reality.