There was a time when calling an act robotic might have been an insult, but these days it’s little more than a descriptive fact. The robot comedy scene is on the rise: there’s Marilyn Monrobot, who was created by Heather Knight at Pittsburgh’s Carnegie-Mellon University and dubbed ‘the world’s first stand-up comedy bot’; RoboThespian, created by the Cognitive Science Research Group at Queen Mary University of London, which performed at the Edinburgh festival, and Glee Club star RoboJase, an android which is modelled ‒ slightly spookily ‒ on The Gadget Show presenter Jason Bradbury. Naturally, robot comedy is big in Japan, with comedian Zenjiro taking a modified PaPeRo (Partner-Type Personal Robot) on tour with him, and roboticists at Konan University developing Gonta and Aichan, who riff to each other about the news.
In terms of robotics these are exciting times, but in terms of comedy, the obvious question is… why?
There is serious science behind all the fun: the origin of robot comics lies in research aimed at exploring what types of interactions and characteristics robots should have to allow people to create a mental model of what the machine is or does. This can be done with standalone signifiers ‒ the humanoid machine slumped in a corner with hunched shoulders is more likely to elicit pitying responses, for example – but more intriguingly, advances in sensor technology and pattern-recognition algorithms for big data now allow robots to perform ‘non-routine’ tasks, including visual and language recognition. What this ultimately means is that robots can modify their behaviour in response to external cues. The potential that these tools offer for ‘embodied’ (rather than calculating) intelligence is a source of ongoing discovery ‒ leading to creations as varied as roboticist Guy Hoffman’s adorable machines that can jam on the marimba and bob their heads in time to the music.
But humour is complex. It can be prompted by verbal, physical, visual, even aural cues. It calls for laser-sharp timing, as well as the ability to read and react to the mood of an audience. RoboJase, PaPeRo, Data and RoboThespian all come equipped with a variety of visual recognition technologies and audio processing tools with which to analyse audience responses and adjust behaviour accordingly. Data and RoboThespian were built to explore the nature of live experiences and comedic interactions. But there’s still a long way to go ‒ so far, all of our comic bots perform work written by professional comedians and presenters rather than creating their own jokes.
But the reality of bots creating their own material isn’t as remote as it may seem. Away from stand up, bots are flexing their muscles over on Twitter. Some bots, such as @Betelgeuse_3 (which replies to any tweet containing “Beetlejuice! Beetlejuice! Beetlejuice!” with “IT’S SHOWTIME!”) are purely reactive, but there’s sharper satire on offer, too. @TwoHeadlines tweets automated jokes created by splicing different newspaper headlines together. Gems so far include “Elon Musk returns, with more blood, revenge and a feisty makeover”. As its creator Darius Kazemi says, the jokes aren’t just about current events, but take on a “very specific future dictated by what a Google algorithm believes is important.”
Algorithms creating biting satire about digital culture may yet be the future of artificially intelligent comedy.
Georgina Voss co-hosts the Gin and Innovation podcast. Follow her on @gininnovation or @gsvoss. Image credit Andrew B Meyers.