uncanny valley

Terminators and the Futility of War

 Terminators. I love 'em. On screen at least (even the much maligned Terminator Salvation). And those of you with an interest in such matters will know that there is an increasing trend in the military forces of the developed world to weaponise machines so that they can do the fighting instead of human soldiers. 

So we are gradually moving towards robotised armies and with that, comes a debate as to whether machines should ever be empowered to decide for themselves whether to kill a human being. Currently, that decision is beyond the ability of machine systems - deciding whether an object is human or not is a step that is still being tackled - but in time, no doubt the problem will be solved.

Each month drone attacks by Predator strike craft take place in Afghanistan, Iraq, Yemen or Pakistan, killing alleged terrorists and their allies. Often civilians are killed in these attacks too and the incidents take up few column inches in reports by the Western media. There is no outcry, no mass protests, no campaign against these attacks. And when we think about the future of war, we need to think about that which goes along with it - public opinion - shaped by what is reported in newspapers, on the internet and TV. War does not survive without the propaganda to support it.

If war is to be waged increasingly with automated craft like the Predator drones and their robotic successors, then what will that do to the perception of military campaigns? There is huge public concern in the UK over the deaths of around 200 soldiers who have been lost since the Afghan campaign started in 2002, but for an operation which began eight years ago these losses are relatively light in military terms, considering the length of the campaign.

So if we fast forward to 2025, when the US has achieved Full Spectrum Dominance and has in place a more robotic army - well within the scope of its military forward planning by the way - what then for our perception of war? If Allied casualties remain light in future wars - then who is going to oppose distant operations against people who are perceived to be terrorists - whether or not they are in reality. And what of the innocent civilians inevitably killed and injured by remote drones?

War, whether fought by humans or machines, is an anachronism, a lesson in stupidity and failure. Each year in the UK, on the 11th November, we have Armistice Day to remember those lost in various wars for 'freedom' against whoever happens to have been the enemy at the time. Solemn ceremonies are repeated year after year, public silences are held and then the world turns and goes back to the same practices that lead to wars in the first place. Lies are told by governments, patriotism is sold to naive young people, lives are given, the media huffs and puffs, atrocities take place at home and in distant lands, body bags are filled and nothing - ever - is learned.

So the debate about whether machines should be enabled to kill is, in a way, the wrong debate. The debate needs to be about how we end war, how we end terrorism as a weapon of the dispossessed, how we cooperate with those who are different to us, how we share precious resources.

One planet and like it or not, we have to share it.

New Story Synchronicities

According to Wired.co.uk, some mad scientist seeks an astronaut to make a one-way trip to Titan. Apparently the ad specifies: “someone unique that has always wanted to see the universe first-hand and has perhaps a terminal view on life here at home." Well, that's one way of putting it. More.

And the lovely False Maria, the robot from the 1927 film Metropolis, is on display in Paris. In case you can't get there, go here instead.

Explanations of the connections between these obscure links soon enough.

Concorde, Cold War Modern & Japan Car

Three exhibitions this year have proved quite revealing in what they have to say about global attitudes towards the future. Most recently, I caught the Concorde exhibit at Manchester Airport. The Great Bird is parked in a hangar, elegant and poised. All dressed up but with nowhere to go. Rather than criss-crossing the world as it is still capable of doing, the plane is grounded by that good ol' one-two punch - bad economics and lack of imagination. Concorde's future is lost, for now at least.

A quite different experience with the brilliant Cold War Modern show at the V&A in London at the turn of the year. That show looked at the period from 1945-70 from a design perspective, detailing some amazing products and structures - from the mundane to the awe-inspiring - as well as a whole host of artefacts that never came to be. Cyborg technology, immense Soviet architecture, Buckminster Fuller's domed cities and space stations all shone briefly on the drawing board and then came to nothing. It was a glimpse into other worlds - one that I grew up in, and one that might have been, both shaped by the splitting of the atom.

Japan Car at London's Science Musuem was a look at more likely futures - one where cars are either inspired by the kei car concept or are capable of zero emissions. It featured the Honda FCX Clarity, the unbelievably kawaii Toyota IQ and some interesting GPS-data-as-art from Tokyo.

Past trips to the US revealed that Cog and Kismet, two innovative social robot projects from MIT in the 90s and early 2000s, are already exhibited at the MIT Museum just outside Boston. At times it seems as if our future utopias are really destined to be just future museum exhibits. Relics of eras that falter once past the concept stage. One can easily imagine that we might find ourselves wandering through exhibits in 20 or 30 years, marvelling at Asimo or the Repliee androids, contemplating another world that might have been.

Concorde

Toyota iQ

Toyota iReal

Tomorrow beckons, fragments of utopia in a long winter.

HRP4C Gynoid Robot

There was startling footage this week of the HRP4C Robot Girl on show in Tokyo. One of the applications the robot is designed for is to act as a model - a moving mannequin - and it will take part in a Tokyo fashion show next week. You can see from the footage that there are still challenges ahead for the designers - Japan's National Institute of Advanced Industrial Science and Technology. The inert hands, the 'old woman' walk, the slow response times for facial expressions. However, the designers say this version is just the first step.

It's interesting that this robot is being considered for work in the fashion world. A few years ago, an American university team put out a call for an artist to help them design the head of a humanoid robot. I don't know what became of the project, but it strikes me that it's now time for artists and designers to work with roboticists as we move closer and closer to robots / androids that will be present in our daily lives. The famous Japanese artist Hajime Sorayama worked on the design of the first generation of the Sony Aibo and so successful was the result that the Aibo is now in the permanent collection of the Museum of Modern Art in New York.

This HRP4C is in part a futuristic body with the head of a rather anodyne Japanese girl. Its hands don't move but are detailed with pink nail polish. In appearance, it is neither one thing nor another. There seems to be a tendency for androids - such as Repliee and Aiko - to look and dress quite conservatively and as a result they resemble awkward Sunday school teachers. We live in a post iPod world now where technology and design are inseparable - if these human-like creations are to feel comfortable, familiar and desireable to us, the makers need to reflect contemporary styles in both their design and appearance. Otherwise, the view that these (mainly) female humanoids are mere curiosities or demure surrogates for the inadequate will persist.

iCub Comes To Manchester

It's rare that advanced humanoid robots make it to Manchester, so when child-sized robot iCub was in town yesterday I made sure to get a place at the evening talk. And a fascinating session it was too, as project leader Professor Darwin Caldwell ran through the robot's development and capabilities. He revealed that they'd given it a cartoon-ish face to avoid the dreaded uncanny valley effect (though they'd forgotten to bring the face with them); that there are around 20 iCubs in existence; that his view is that the domestic robot will appear between 2020 and 2030 (nice tie in with Exilium then) and that European robotics has the edge on the Japanese in a few areas, such as mechanics and cognition. The robot is impressive and reminded me of another child robot, the CB2, developed in Osaka.

Professor Caldwell added that many technicians working on the Japanese robots don't actually know how the whole device works, just their part of it so it was interesting to discover that iCub is an open source project - anyone anywhere can contribute to it and if you've a spare £200K to build your own iCub, you can download the software into your home version and it will crawl around for you.

Everything iCub is here and there is video footage at the BBC news site.

The Mystery of Embodiment

This is cool. Check out this Times story about 'body-swapping' - basically using sensory input to fool the mind into thinking that it inhabits a different body. This is something that Hiroshi Ishiguro said they'd found when experimenting with his robotic double, the Geminoid. Whoever controlled it could 'feel' a sensation of being touched whenever the robot was touched, even though the robot was some distance away. More on that at my main site.

Read. Be amazed. And let the philosophical wonderings commence.

Mirror Neurons & The Uncanny Valley

...'I don't know what I'd do without you.'

'Mirror neuronal arrays. Synthesised endocrine system. You can't help but feel that way. Nor can I.'

She frowned at him.

'They made us just for each other,' he shrugged, 'and we're caught in an existential riddle.'

*    *    *

The relationship between Alex and Kim in Exilium is partly explained through their inbuilt 'mirror neuronal arrays'. Kim's experience of empathy also comes to down to this aspect of her neural architecture. The wonderful robots.net website features a link to a paper (PDF format) by Dr. John R. Skoyles, explaining that in humans mirror neurons help 'blur the physical boundaries between ourselves and others'.

It's a fascinating area, with much that is relevant to robotics. Mirror neurons may even explain some elements of the uncanny valley phenomenon, why inexpressive 'mask-like' faces are so off-putting to us. It's not just that robots will need to experience empathy towards us and each other, but that they will need human-like behaviours if we are to feel comfortable in their presence.

The New Laws of Robotics

About five years ago, Japanese roboticists proposed the Atom project - a thirty-year plan to develop a learning humanoid robot with intelligence and sentience equivalent to that of a five-year old child (for background see this link). The proposal was compared in scale to the USA's Apollo project in the 1960s and it was envisioned that Atom would stimulate the economy and provide multiple spin-off developments in new technologies, just as the space programme had done.

At this stage, Atom remains just a proposal - although the Japanese government does fund many smaller-scale robot projects - but what if one day it does get approved? What do we want our robots to do? What do we want them to be? Much has happened since Isaac Asimov proposed his legendary three laws and as robotics constantly evolves, perhaps it's time to look anew at the issue of robot ethics. We need to look at what qualifies as intelligence, how we determine emotion and whether we want to give robots the ability to kill human beings, whether they will need protection from our worst excesses.

A discussion like this can't help but reference the work of Isaac Asimov (as well as Lyuben Dilov and Nikola Kesarovski), so with that in mind what guidelines are needed for the next generation of robots - and the humans that make them?

1: A robot should represent the best in human nature - exemplifying compassion, self-sacrifice and empathy through roles as carers, therapists, explorers, co-workers, companions, teachers or entertainers.

2: A robot must protect human and animal life.

3: A robot must protect its own life, unless laws one and two require otherwise.

4: A robot need not obey instructions given by a human being if such instructions conflict with laws one, two or three.

5: A robot must be committed to its designated tasks, unless this commitment conflicts with laws one, two or three.

6: A robot may not be subject to abuse or degradation by human beings.

7: Sentient robots and androids have the right to recognition and protection as individuals under national and international laws.

8: If a robot has human form, it must know it is a robot and declare its status when asked.

...Discuss.

Will Artificial Intelligence Help Journalists Do Research?

A recent piece in the UK's Guardian newspaper by Charles Arthur gives a good example of a) poor journalism and b) ignorance of what artificial intelligence researchers are actually trying to achieve. The original article is here, but if you'd rather read it later then allow me to summarise - the piece is entitled Artificial Intelligence: God Help Us If Machines Ever Think Like People and in it, the writer argues that:

Machine intelligence shouldn't be modelled on the human mind because we are inherently flawed beings, caught in an eternal battle between logic and instinct. Any machine with a brain modelled on the human mind will thus end up neurotic and ineffectual.

The piece shows little understanding of why AI researchers are trying to create machines and systems that possess high speed sensory processing, vast memory reserves, the ability to adapt to fluid environments, objection recognition in a variety of locations from small amounts of basic data, adaptability, language, sense of self and so on. There are a whole range of things which we humans are quite brilliant at and which any robot should be endowed with in order for it to be able to function successfully in our world. So why wouldn't scientists want to model their machines on proven biological models?

Articles of the God Help Us... variety often appear anytime someone or something pops up that is left of field or is in some way visionary. When mainstream outlets like The Guardian publish these non-think pieces, one could be forgiven for concluding that their goal is not to inform but actually to permeate a state of ill-informed, post-modern ennui.

 

(originally published 3rd July 2008)

 

Dr Ian Hocking said:

Hey Richard - Just thought I'd pop by and say hello. I read 'Robophobia' a couple of years back and enjoyed it. Feel free to pop around to my blog (http://ianhocking.com) if you fancy an interview on the blog to publicise your new book...

Cheers
Ian

 

(15th July 2008)

What's Inside Asimo?

Honda have revealed a new networked version of ASIMO this week - it has obstacle avoidance and other new skills that bring it ever closer to being a useful companion. Elsewhere, one of ASIMO's technicians, Ekkasit Najaitrvek, is quoted as stating that the robot is 'much more than a super-computer'.

"He's a person," Najaitrvek says. "Sometimes he's like a kid. Most of the time he's my friend. "You can see when he looks at you -- you can't deny he's a person."

When I saw ASIMO at the London Science Museum in 2004 (full story here), I had a similar reaction - at one point it 'looked' right at me and I felt I was being acknowledged by a person. It was an innate, unexpected response. Why is it that ASIMO - and other robots - have this effect on us? After all, ASIMO is only at the start of its independent intelligence - it is an infant mind, more programmed than autonomous. Clearly appearance, movement and purpose add up to something greater than their individual parts - the question is, what is it that they add up to?

(originally published 16th Dec 2007)