Researchers at MIT have used AI to create a chatbot that can talk to your “future self,” reducing anxiety and demotivation in users after the chat.
Have you ever dreamed of talking to your future self when you're 60 years old?
A team of researchers from the Massachusetts Institute of Technology (MIT) in the US may offer the closest we can get to experiencing that.
To do this, they use artificial intelligence (AI) to mimic natural conversation.
The researchers used GPT-3.5, a popular model developed by OpenAI, to ask study participants a series of questions about their lives, past experiences, and aspirations.
The system then created a “user backstory,” which the study authors also call a “synthetic memory.”
For example, a user may aspire to be a high school biology teacher.
The AI future-self chatbot can recall “memorable moments” from when it took students on a “field trip to a local nature preserve.”
“Future self-continuity”
The MIT researchers exploited a behavioral science concept called “future self-continuity,” which suggests that people who feel a stronger connection between their present and future selves are more likely to make decisions that lead to their long-term well-being, including financial, educational and health decisions.
“Humans often struggle to vividly imagine their future. This limitation leads to a cognitive bias called 'temporal discounting', which causes people to prioritise short-term benefits over long-term ones,” Ivo Vlaev, professor of behavioural science at the University of Warwick in the UK, told Euronews Next in an email.
The chatbot, created by MIT researchers, included a profile picture that was altered to show participants themselves as 60 years old, helping to reduce time discounting.
“When people interact with their digital selves, the future feels more tangible and close, reducing the psychological distance between their present and future selves,” Vlaev added.
For the professor, chatbots embody the idea of nudges, or subtle interventions designed to guide behavior in beneficial ways, guiding users to make better decisions by making their future selves more relevant to their present selves.
Improved health
in Pre-printed paper Although it has not yet been peer-reviewed, the MIT team tested the app on 344 people between the ages of 18 and 30.
The researchers found that after speaking to their older digital selves, “users reported reduced anxiety” as well as reduced “negative affect” and lack of motivation.
More research is needed to learn more about the results, but the effect may depend on whether the chatbot sounds lifelike, Vlaev said.
“Using digital throwback images is a clever idea because imagining yourself in the future adds a layer of continuity and realism,” he said.
“But it's not clear whether people did anything differently or changed their behavior as a result of interacting with their future selves,” Vlaev said.
In a section on ethical considerations of the study, the researchers noted the need to be mindful of AI-generated content that may encourage negative or harmful behavior.
