It's replicant Roy Batty's birthday – but hey, where's my killer robot?
Blade Runner got robotics so wrong
Comment In the Blade Runner universe the Nexus 6 replicant calling itself Roy Batty rolled off the production lines of the Tyrell Corporation today. Sadly, or some might say luckily, the tech industry hasn't yet caught up with Hollywood.
I do have a Nexus 6 on my desk right now, but it's a sizable phablet of a phone, rather than a humanoid killing machine. Technology still has a long way to come.
Batty, masterfully played by Dutch actor Rutger Hauer, was what screenwriters thought would be coming down the line a generation later than filming. Like most science fiction creations, Batty's timeline was horribly wrong, but why don't we have humanoid robots walking among us these days?
Firstly, programming a robot to mimic human beings is incredibly hard. The bipedal locomotion we're familiar with is something humans have evolved over a few hundred thousand years and it turns out it's surprisingly tricky to replicate.
Balancing a high torso on mechanical hips and making a robot walk isn't something that's easy without a lot of computing power to stay stable. Anyone who has got wankered on a Friday night will appreciate the problem – balance is one of the first things to go when the brain gets sloppy.
DARPA is currently trying to find a humanoid robot that can handle even simple tasks like opening a door or driving a car. So far the results haven't been good, to say the least. Building the hardware is easy – getting the motor skills to move it is something else entirely.
At the time Blade Runner was made, mainstream computing was in its infancy. IBM had just brought out its first PC, chip manufacturers weren't even close to gigahertz speeds, and the rest of computer hardware like memory and hard drives were (comparatively) huge, slow, and very expensive.
Batty was listed as a high-performance combat drone, with excellent intellectual abilities that would have put it in genius mode. Leaving aside the physical aspects of robotics (and clothing them in human skin – – which could well take another hundred years to manage) there's still a considerable brain gap to fill.
After decades of trying, we're still nowhere close to artificial intelligence or even a computer that could pass the Turing test or convince JF Sebastian to trust it with his life. Even the most advanced machine-learning systems can be beaten by a five-year-old incentivized by a chocolate bar.
Getting silicon smarter takes huge amounts of computing power – server racks full of hardware and carefully designed software subroutines. Cramming that inside a human-shaped body isn't going to happen for a long time.
In short, don't expect humanoid robots that can mimic their creators in the way Ridley Scott envisaged in our lifetimes, or even in those of our children. This isn't necessarily a bad thing however.
The fact is that humans are far more efficient at doing humanoid functions than a robot will ever be. The future of robots is in designs that serve our needs, rather than trying to replace us.
As a case in point, look at Batty's dying speech. Written by a human brain, and improvised by the actor, it's still one of the most moving soliloquies yet created – no automation required. ®