Honesty with Machines
I watched a TED Talk entitled "Why we are more honest with machines than people," by Anne Scherer. She began by talking about a research team who looked at the data of US soldiers post-combat. Soldiers were required to talk about their health with a healthcare professional, but the researchers suspected that the soldiers weren't openly sharing their concerns for fear of being perceived as weak. The team developed a chatbot that could interview the soldiers, and they found that during digital interviews soldiers reported more health problems than when speaking with a human. Removing humanized interactions allowed the soldiers to be more honest in their reports. With this information, Scherer's research team wondered whether creating more humanlike bots would make people less likely to be honest. Her team developed two chatbots—humanlike and machinelike. The bots asked separate groups a series of questions that got increasingly more intimate. The results showed that the humanlike bots yielded more socially desirable responses, while the machinelike bots yielded less socially desirable responses. She concluded by stating that dehumanized interactions may prove useful if honesty is the goal.[1]
I found Scherer's talk to be interesting and refreshing. There is a lot of coverage on how machine interactions are negatively impacting humanity, but Scherer showed us something different—machine interactions could enable more honest responses, particularly for vulnerable populations like veterans. I also appreciated that her research team asked how things might be different with a more humanlike bot.
Scherer is an Assistant Professor of Quantitative Marketing at the University of Zurich with interests in the psychology of technology and machine influence. According to her website, she relies primarily on online research and secondary consumer data for her research.[2] Her work in behavior psychology, perception, motivation, and social psychology is quite fascinating. I would have liked to dive more into her research specifically with the chatbots, but it was not cited in the TED Talk, footnotes, nor on her website.
Her work provokes interesting implications for our relationship with technology and each other. Though she does research geared toward business and marketing, I'm curious about the social consequences. If I can be more vulnerable with a machine, could I develop an intimate relationship with machine intelligence(s)? Could machine intelligence(s) play a role in therapy due to the benefits of non-judgmental responses? What would be the pros and cons of a humanlike or machinelike bot chatting with therapy clients? What would this mean for human-to-human interactions? With the advancement of machine intelligence(s) and our increased interactions with them, there's plenty of opportunity for research into human engagement with machines.
Notes and Citations
[1] Scherer, A. (2021, July). Why we are more honest with machines than people. Anne Scherer: Why we're more honest with machines than people | TED Talk. TED Talk. Retrieved May 15, 2023, from https://www.ted.com/talks/anne_scherer_why_we_re_more_honest_with_machines_than_people
[2] Scherer, A. (2012, June). Short Profile. Anne Scherer. https://www.annescherer.me/home/short-profile
