We talk Van Gogh, artistic AI and the philosophy of technology with creative robot developer Paul O’Dowd, ahead of his appearance at the Affordable Art Fair

So Paul…why are you fascinated by the idea of a robot that can draw?

My fascination is actually with the process of trying to create robots that can draw, and do other things. There is a lot about being human that we learn by interacting with the world, and then find difficult to put into words. Children take to drawing and painting naturally and with ease. But as soon as you try to program a robot to work in a similar way, it exposes huge amounts of complexity. How do we hold materials? How do we know the difference between pencil and charcoal? How do we decide how fast to move, or how much pressure to use? And how are these things used when we want to express ourselves? These sort of questions are very difficult to answer in words, but with a robot I can explore these questions in a proactive, practice based way. More often than not, I find myself amazed at what humans can do. 

“…working with robots has always been a way to ask philosophical questions about humanity, about intelligence and about the world around us…”

For me, working with robots has always been a way to ask philosophical questions about humanity, about intelligence and about the world around us.  From the youngest age I have been puzzled by how the natural world seems to know what to do without any kind of instruction manual. Robots are the exact opposite, everything needs to be considered – from how to physically build one, to how to program them to make decisions. So the difficulty in creating robots becomes a philosophical tool; I try to imagine what a robot might see, and how a robot might create meaningful behaviours. It makes me question how I do things that I otherwise take for granted. And what’s more, putting a robot into the public is a great provocation, it creates a huge amount of healthy conversation about the future of technology.

Paul O'DowdPaul’s robot at work

What are the biggest challenges when designing a robot of this kind?

The biggest challenge is that technology is often designed to solve an exact problem. Hammers for hitting nails, blenders for chopping food, calculators for doing sums, cars for travelling. But Art is not exact. Robotic technology in particular has developed in the manufacturing industry, where it is a tool designed for precision and predictability. The Arts cannot be described as a predictable and neat function. Instead, we value all of the nuances, contradictions and idiosyncrasies of what it means to be human. The Computer Arts and Digital Arts has had a hard time being recognised as Art. I suspect that this is because the art work is saturated or highly mediated with the logical nature of the underlying medium of digital technology. But with robots, there is an inherent analogue quality – they are physical machines that wobble, or sense things badly, or make bad decisions. When we say ‘dance like a robot’ (i.e., to be stiff) this has more to do with how it was programmed than its physical character. So the hard thing is to use a highly logical technology to describe and express subjective qualities. In this way, robots stop being a tool, and become a medium like paint. I almost have to forget what I have been taught, and instead push the technology beyond its original design.

Do you have to tell the robot what design to create? If so, how to you go about this?

Yes and no. My education has specialised towards investigating emergent intelligence. A good example is a flock of birds. They fly as a group, and no single bird is in control. And yet they stick together and avoid crashing into trees and buildings. We can’t say any one bird has the intelligence responsible for the group. Instead the intelligent behaviours we see in a flock are a result of interactions. I think a very similar process underlies art. An artist has an intention, but the paint or charcoal behaves unpredictably, so the expression becomes shaped by that moment of interaction between person and world.

Paul O'Dowd

A jellyfish drawn by Paul’s robot

The robot has a similar emergent quality. Currently, I provide the robot with a digital image as a foundation, such as a photograph. A photograph has a lot of information in it. I’ve then written a series of algorithms. The first algorithm attempts to find texture in the image, which creates new data on how lines flow in the image. The second algorithm then tries to join up the flow lines, which works a bit like running water over the image, finding ‘natural paths’. And the last algorithm then simplifies all the lines, often removing 90% of the original data,  which gives them curvature and speed changes which will affect the way the machine physically draws with the pen.

At the moment it is quite a human-robot collaboration. When I give the robot an image I can only guess at what will come out. So I have written an interface that lets me explore the way it has been processed, to try to understand the information, and create an image that best expresses the characteristics of the technology.  

What do you see as the future of AI? Are we about to be overrun by robots or are things slightly more nuanced than that?

I think we should all be discussing AI as much as possible. AI can be generalised as very powerful pattern recognition technology. There have been recent press articles about how AI has been used to target political campaigns at vulnerable persons, called Dark Ad’s. Obviously, this is a very serious issue, and a consequence of AI right now, not in the future. 

Paul O'Dowd

An example of Asemic (wordless) robot writing

However there is an important distinction between AI and robotics. Most AI technology runs solely inside of computers, and therefore does lots of number crunching. However, robots work in a world which can’t be entirely described by numbers (such as, a number value for joy), and robot bodies are still very unsophisticated (for example, there is no robot equivalent for a touch sensitive skin). The effect of this is that robots are still very clumsy. Just trying to get a robot to understand something like drawing with a pencil is exceptionally difficult!  We can make robots do very exact, well-described things. So we should also be discussing robotic military drones that drop bombs, for example.  However in my opinion, robots that can fully integrate in our complex and messy world, in a science-fiction sense, are still a long way off. 

Another important subject is robots taking jobs, which is a very alarming phrase. However technological progression has a long history of displacement which our society has adapted to. The bigger issue is who owns the technology.  If only a minority of people have control of robots in the economy, then only the minority will stand to benefit, and the majority will suffer the displacement. So like our current problem of the distribution of wealth in our global society, robotics presents a problem of a distribution of empowerment that could make things much worse. In my opinion, robots that can do exact tasks and tedious work are a good thing, but only if their integration in our society is well-considered. There are discussions of solutions like a robot-tax to help balance the scales. 

How did you get into creative design and engineering as a profession?

I was one of those annoying children that continually asked questions. When the answers dried up I needed to find other means. Bristol has a vibrant community of people asking questions and practicing ways to explore them. I came to Bristol to study robotics and since then I have done my best to stay immersed in good conversations. To me, creativity is as much a part of science or engineering, as it is in art and design. However, it is the mixing of disciplines which is key to gaining new and valuable perspectives. A specific training can be restrictive, so creativity is about mixing things up. A key moment for me was co-founding the robotics art & design studio Rusty Squid Ltd with David McGoran and Roseanne Wakely. The studio is another means to ask questions and continue conversations with as many people as possible. My degree was a bachelor of science, so starting an art & design studio was a direct way to transpose my scientific training into unknown territories. There hasn’t been a plan, I’m still just asking questions. 

Affordable Art Fair visitors will be able to see the robot in action, creating works like this

Do you have any advice for students hoping to develop their own AI?

AI is really exciting at the moment because the technology has become much more accessible. A lot of AI technologies are open source, and there is a wealth of free education available on the internet.  Simple determination will make good on the available technology and resources. My advice would be to nurture one’s own creativity as much as possible. AI is a toolbox, and to use it we have to be able to make useful observations of the world around us. So keep sketchbooks, read lots, write a journal, seek conversation, and pursue diverse influences from experiences had in the world. 

What can Affordable Art Fair visitors expect from your robot when they come to the fair?

The robot itself is very pragmatic looking, not humanoid, not even a robot arm. In a sense it is not very theatrical. However I think this is important. It makes it harder to project ourselves on to the way the machine is working. And so it makes it easier to see how difficult it is to program a robot to draw, and how incredible it is that we can draw so naturally. The robot is also quite alien, it moves very quickly and makes lots of squeaking noises. I think this helps to have a conversation about what robotic technology can add, what unique characteristics it has that can add value. We could build robots to look human, and we can program robots to copy a human style of drawing, but there seems little value in doing so, because robots are fundamentally not human. A robot copying a human is disingenuous, underwhelming. 

“…for a robotic artwork to become revered like a Van Gogh, it will have to transcend our expectations and notions of imitation…”

Do you think work created by a painting robot will ever be revered in the same way as, for example, a Van Gogh piece?

Yes, I think it is possible. I think a significant value in art is the perception of risk, whether that is an intellectual risk taken by the artist, or the risk of spending so much of one’s lifetime to develop an unprecedented level of skill. For example, Fountain by Duchamp was controversial, but imitations are worthless. In a similar way, for a robotic artwork to become revered like a Van Gogh, it will have to transcend our expectations and notions of imitation – it will become an original contribution, a testament to human creativity, aesthetics and the sublime.

What exciting projects are you working on at the moment?

My most exciting project is definitely Rusty Squid Ltd. It is a project that will probably exist beyond my lifetime. It may even evolve into studios or businesses of other names. Rusty Squid has become a perpetual context for me, whether I am employed by the studio full-time or employed elsewhere in other contexts. Starting a business with Rosie and David solidified our joint philosophical enquiry, and has started an inexhaustible conversation on technology and humanity. I feel like all my activities simply feed into the community discussion growing up around Rusty Squid Ltd.   

What is your biggest ambition in terms of technology and design?

This is an interesting question because I have not really thought about it in this way before. I haven’t really ever set myself an end goal, I’ve always been so engaged in the pursuit of ideas. However,  recently I’ve started to realise how philosophy is actually embedded in the digital technology we design and create. Digital technology is built upon 0’s and 1’s, and really, the belief that any problem can be simplified into this numerical basis. I’m just waking up to the fact that technologists are educated to reduce problems down in this way, and necessarily so, because digital technology and the world are very complex.

However, that means digital technology is designed by people who are philosophically reductionists, and the people who use digital technology are only presented with reductionist ways of interacting. For instance, we are all used to icons to press, simple icons. No wonder some people feel alienated by technology! I think this might have a profound consequence on how we perceive technology, and what we generally think is possible. I’d like to challenge this, and to do this by exhibiting robotic artwork that proves a new philosophical approach to technology. 

What have you got coming up for 2018?

I’m just finishing my employment contract with the University of the West of England, which has been a really stimulating period for me. I’ve been a researcher and tutor there. Often times I felt like the students, with their challenging questions, were teaching me so much about art and design! I’ve always been very driven to find the next thing, so my new challenge is to actually not have a clear next step, and to see what happens next. 

If you would like to see Paul O’Dowd and his artistic robot at the Affordable Art Fair, which takes place from 8 – 10 September, visit affordableartfair.com to buy tickets and find out more