Rebecca Lynch, Visual Storyteller in Residence at KCL
by Jennie Williams & Zhouyue Xia
Rebecca Lynch, Faculty of Natural & Mathematical Science supported, creative writer-in-residence at King’s, is currently working in collaboration with Professor Elizabeth Skylar, Head of the Centre for Robotics Research. Rebecca was invited to CUSP London as one of the guest lecturers on Telling Stories with Data, a module in the MSc in Urban Informatics, to talk about storytelling and data.
After a warm welcome from Dr Alfie Abdul-Rahman, module lead of Telling Stories with Data, Urban Informatics students, lecturers and guests, Rebecca began her lecture.
The Eternal Cycle of Spectacle and Storytelling.
Rebecca, the artist, introduced her avatar (Fig.1) to set the scene:
“I don’t really have an interest in creating things that are beautiful but have no meaning, I’d rather create something that’s … a bit rough, but has a good story”
Rebecca was briefed to talk about data and storytelling and, interestingly, decided to interpret that freely, which made for a thought-provoking and imaginative lecture.
“When we look at things that involve data and storytelling at the moment, a lot of time we seem to get caught up in terms of what we can do with technology and how we can make something look really good and look beautiful and we forget to ask, actually ask ourselves what it is we are trying to say”.
In the 19th century, 6 million people attended the Crystal Palace Exhibition of 1851, a showground spectacle to demonstrate new ideas. Showmen were perceived to be demonstrating magical things, e.g. oxygen, capturing the audience with the spark ignited in the lag between understanding and amazement. An example of spectacle is P.T. Barnum, a showman in the 1800s, infamous for many curiosities but most of all his circus, spoofs and scientific entertainment.
“Wherever there is technology, historically there has always been spectacle”
Spectacles today include Artificial Intelligence (AI) demonstrations using goggles, connecting an audience with technology whilst storytelling. Rebecca’s talk wowed with the modern day spectacle; the virtual whale 7d video by Magic Leap, a $4.5 billion virtual reality company.
“The problem with this video was, is it real or fake?….. it’s a sort of visualisation of what
their product will do in the near future…….. we are constantly slightly projecting ourselves
into the future, all the time”
Fake being the current buzzword for the last couple of years, perhaps replacing C19th magic in the C21st? Hollywood was mentioned, its big-budget films working around technology, the sophistication of technology outrunning realistic storytelling.
“often sacrifice the plot, they sacrifice the story and they sacrifice the character development”
“An example of focusing on the technology, rather than what we can do with it” is DeepFake: a highly technical platform involving creating fake videos, in particular, of celebrities. The technology involved combines deep learning with images to create a fake persona, naturally focusing on the technology rather than storytelling.
In recent years, data visualisation has been quite the striking visual spectacle itself, graphics trying to portray a visual story, caught up in technology and detracting from what we are trying to say. Around 5 years ago, Dave McCandless published “Information is Beautiful”, a book full of spectacular graphics, some very difficult to interpret or understand; raising the question: does it matter what they are trying to say, as long as the graphic looks good?
Rebecca described this as “the error of peak visualisations” proving her point by finding an interactive graphic from a random website and making up a completely fake story of what it represented.
“Human beings are not always that good at interpreting things visually, actually, and we can be fooled and we can be led by visualisation.”
So, are we progressing from the ‘spectacle’ to the data visualisation stage? The spectacle stage is useful; it drives inspiration. Rebecca then presented some ‘good’ examples, info- graphics, which reveal stories behind the data and are not just a pretty picture:
A simple graphic with a strong message:
the Bechdel Test: detecting female characters in books/movie scripts:
These sources produce interesting visualisations, worth exploring later.
Rebecca demonstrated a global temperature DataMap, an interactive graphic with the dimension of time, alongside a few simple but effective story-telling graphics.
“So, with storytelling, the key factors are, to be successful in a story, are relatability…….and empathy…”
“What makes a good story……What makes a good story with data?”
Let’s look at the relationship between data and the story:
|Data is…||Story are…|
|=TRUE||True, False, Fake or Fiction?|
Quoting Joan Didion, an iconic American writer of the 1960s & 1970s, Rebecca reads:
“We tell ourselves stories in order to live…We look for the sermon in the suicide, for the social or moral lesson in the murder of five. We interpret what we see, select the most workable of the multiple choices” – Joan Didion, The White Album.
A type of story, a fable, is usually a story which tells us what to do and what not to do, a story letting us know the consequences of our actions, similar to:
The “dangers of technology and how we might react, or should react, when we find ourselves in various situations regarding technology…”
Rebecca currently has an exhibit in the Bush House Arcade, available to experience until 15th December, 2018 and she wrote an environmental fable, visualised in the exhibition, where:
“women are followed by personal surveillance drones for their own protection, and it’s a trade-off between personal freedom and being protected or secure”
Human stories are equivalent to training data in AI, and in the 1980’s Kurt Vonnegut & David Yang introduced the shape of stories.
Moving back to stories and data…
Recently researchers recreated the shapes of stories idea computationally at the University of Vermont. So, stories can be datafied and we can use stories as training sets in AI but:
“Artificial intelligence is, in a way, leading us down a path which carries along with it all of our historical biases”
Artificial Intelligence (AI), are machines that can be trained to learn, solve problems, make decisions, even when facing a new scenario, and attempt reasoning. AI is coded using neural networks, which mimic the human brain. Tools available to perform such tasks include Googles Tensor Flow, an open source library using data flow graphs on node/edge data, mapping the flow between them. Google Inception, another AI tool, which could be thought of as “hallucinations of AI”. Rebecca played a clip from Deep Dreaming Fear & Loathing in Las Vegas: The Great San Francisco Acid Wave, to highlight the extremes of AI capabilities.
We are at the stage now where AI/Neural Networks could be told a story or fed a script and produce a film based on language and its image training set.
Rebecca concluded her lecture exploring the idea of whether we could mimic creativity with AI, teaching a computer to draw images by processing thousands of training images. Several AI generated images were displayed, interesting, the computer trying to generate what it was asked to produce, but it’s not quite there yet.
Just 3 weeks after the lecture, a piece of art, generated by AI, sold at Christies for $432,500. (Fig.2)
In the end, Rebecca also shared her thoughts about her newest work Dreaming Machines exhibited at Bush House Arcade, work generated from the error of AI algorithms output during a collaboration of AI researchers. Furthermore, to get the newest update from Rebecca, her personal website is the best place.
Jennie, Rebecca & Zhouyue at CUSP London
Unless stated all quotes are Rebecca Lynch and were taken from the module lecture capture available on Keats.