When Your AI Assistant Misbehaves
Introducing ‘A Week With Wanda,’ a Mozilla Creative Media Awardee
We expect the technology in our everyday lives to make life easier, often without asking whether it can be trusted. Should we always assume our AI voice assistants like Siri and Alexa have our best interests at heart, or share our values?
We already know AI can go wrong in benign ways, like mishearing our commands or accidentally ordering groceries we don’t need. But as AI becomes more pervasive and more powerful, the stakes are getting much higher: AI is already replacing human interactions, accessing our private data, and discriminating against certain groups when it comes to health care and even prison sentences.
This is the central idea behind A Week With Wanda, a web-based simulation exploring the risks of AI. Play the game at https://weekwithwanda.com
Over the course of one week, users will be introduced to Wanda, an AI assistant who attempts to improve their health, wealth, and relationships. But Wanda’s efforts quickly grow deranged: Through humorous online chats, users will learn that Wanda has signed them up for therapy; sold their location data; reported their friends to the police; stalked a potential new best friend; and even “deepfaked” them into a pornographic video. (Wanda’s actions are simulated, not real.)
A Week With Wanda is created by Joe Hall, a London-based creative technologist. The project is supported by a Mozilla Creative Media Award.
Says Hall: “Most people are sleepwalking into the future of AI: unaware of the risks, and unable to shape the technology’s future. A Week With Wanda is a funny yet thought-provoking simulation of what life could be like with an AI going increasingly off the rails. It’s a way for people to better understand the risks of artificial intelligence, and have their say about how it should change.”
Beneath the veneer of humor and hijinks, A Week With Wanda unpacks complex issues, like how artificial intelligence intersects with privacy, autonomy, gender and racial bias.
Mozilla’s Creative Media Awards are part of our mission to support a healthy internet. They fuel the people and projects on the front lines of the internet health movement — from creative technologists in the UK to digital artists in the Netherlands to computer scientists in the United Arab Emirates.
The latest cohort of Awardees uses art and advocacy to shine a light on the AI that influences our everyday lives. AI today is invisible to most of us, yet has an outsized impact: It influences what news we read, who we date, if we’re hired for that dream job, and whether or not we qualify for a loan or parole.
Says Mark Surman, Mozilla’s Executive Director: “Artificial intelligence is increasingly interwoven into our everyday lives. Mozilla’s Creative Media Awards seek to raise awareness about the potential of AI, and ensure the technology is used in a way that makes our lives better rather than worse.”