Having a personal identity stolen can be devastating. In a digital world full of impersonation Facebook profiles and scam artists stealing personal information through phishing tactics, our identities are more at risk than ever before. Even when our personal information is provided to a legitimate organization, recent massive data breaches have leaked millions of personal records into the public domain, leaving it up to account owners to re-secure their accounts.
In many ways, robots bridge the gap between the digital and physical world. Their electrical bodies paired with connected networks create a physical presence in our environment, while also accessing a digital library of instructions and data. With an increasingly large amount of artificially intelligent (A.I.) devices and robots, our society is about to face a new set of scenarios that require a much higher amount of human robot interaction than what was required in the past. What types of identities should these robots hold? Should these robots have the same identity across all people, or cater their expressions to the individuals they are interacting with? I argue that the design of the identities and personalities inside of these assistants will be a critical factor on how humans adopt and integrate these technologies into their personal lives.
To better understand the impact of human identity within robotics, Ehud Sharlin’s CPSC 584 course at the University of Calgary I designed, then ran, a pilot study that involved a participant having an open dialogue with Baxter. As their dialogue progressed, Baxter would slowly begin to take on the identity of the participant, by mentioning similar interests and personal traits in the conversation. Although, to the participant, the robot was seen as being a fully autonomous artificially intelligent entity, a researcher was actually controlling the responses of Baxter behind the scenes following a prewritten script using a Wizard of Oz testing approach.
The participant was a close friend of mine. I used Facebook’s advanced search tool as a primary source of material for the script. The participant’s posts, photos, friends, and other metadata found on their account, had enough information to develop a rough timeline of key life events to discuss. Additionally, I asked friends and family of the participant for supplementary information. In the future, I believe that an artificial intelligence could collect a similar set information from a participant’s social media accounts and curate the conversation in a similar fashion. All questions asked by Baxter in the script were designed to prompt very specific responses from the participant. On the other hand, all responses from Baxter were designed to be generic enough to allow for slight variations in the participant’s answers. The script itself also includes cues to change Baxter’s face and body language. Over the course of the interaction, the face of the robot was designed to slowly resemble the face of the participant.