PICT.IO: Developing a Collaborative Game for Humans and Machines

Frizzera, L., Zamboni, J., Salles, J. (2019) PICT.IO: Developing a Collaborative Game for Humans and Machines. Canadian Games Studies Association. (CGSA). Vancouver, Canada.

The game is available here: http://www.gamepictio.com/

The presence of Artificial Intelligence (AI) in daily activities has increased intensely in recent years. AI systems recommend movies on video streaming platforms, suggest new friendships on social networks, regulate traffic, calculate financial risks, and try to identify threats to national security (O’Neal, 2016). Through smartphones, wearables, and personal assistants like Siri, Cortana, and Alexa, we begin to have a more personal and intimate relationship with these agents. They help us organize our daily tasks, decide what we want to buy, deliver the news we want to read and more.

However, the techniques used in AI applications are hard to understand, often inscribed in complex algorithms locked behind commercial secrecy, and their possible socioeconomic and cultural impacts are not broadly discussed within society (Gillespie, 2014). All too often AI is presented in a polarized and stereotyped fashion: on the one hand, from a threatening perspective that foresees the domination of machines over humans. On the other hand, AI is also approached as a redemptive technology that proposes the solution to various problems of society. 

The goal of our project is to instil curiosity and enthusiasm in the topics of AI, to clarify technical aspects about the state-of-the-art of machine learning and AI in general, and to encourage reflection on the ethical and social consequences of the technologies in our daily lives. To this end, we present AI techniques in a playful way through a game involving humans and machines. While most of the examples of AI gameplay are competitive games where machines play against humans, such as strategic games (e.g., Chess and Go) and in many first-person shooter videogames, our project has a different, cooperative approach.

Inspired by the game Pictionary, PICT.IO was built using Quick, Draw! API to be used as an educational tool to facilitate AI literacy. Differently than in the usual settings where humans play against machines, in PICT.IO, the machines and humans work together in teams communicating with drawings in order to win the game. Our presentation will report on the 1) development of PICT.IO, including the game design and the challenges to work with machine learning, 2) a quick demo of the game, and 3) the future directions of the project.

Games offer opportunities to learn, solve problems, and become more skilled. That is, indeed, what makes games fun (Gee, 2005). While play as an exploratory method for producing new knowledge breaks with the traditional model of pedagogy practices usually disseminated (Thomas & Brown, 2007), Butler (2012) points out play is our first tool for making sense of the world. Thus, by enabling people to play with AI in a domestic and friendly environment, we are encouraging them to think critically about complex algorithmic processes, and the socioeconomic and cultural implications AI have within society.