Researchers use AI to let you chat with your future self

MIT researchers used AI to create a chatbot where you can talk with your “future self” leading users to feel less anxious or unmotivated after the chat. Have you ever dreamt of speaking with your future self who has lived to the age of 60? A team of researchers from the Massachusetts Institute of Technology (MIT) in the US might give you the closest possible way to experience that. To do so, they use artificial intelligence (AI) to mimic a natural-sounding conversation. They used the popular model GPT-3.5 developed by the company OpenAI and asked study participants a series of questions about their lives, their past experiences, and their aspirations. The system then created “a backstory of the user” which the study authors also called “a synthetic memory”. For example, a user might aspire to become a high school biology teacher. The AI future-self chatbot can recount a “memorable moment” of taking the students “on a field trip to a local nature preserve”.

‘Future self-continuity’

The MIT researchers tapped into “future self-continuity”, a concept from behavioural science which states that individuals who feel a strong link between their present and future selves are more inclined to make decisions that favour their long-term well-being whether financial, educational or health-related. “Humans often struggle with imagining their future selves vividly. This limitation can lead to a cognitive bias, known as ‘temporal discounting’ which in turn leads people to prioritise immediate rewards over long-term benefits,” Ivo Vlaev, professor of behavioural science, at the University of Warwick in the UK, told Euronews Next in an email. The chatbot created by the MIT researchers, which includes an altered profile picture to present the participant with a version of their 60-year-old self, is able to reduce temporal discounting. “When people interact with a digitally-aged version of themselves, it can make the future feel more tangible and immediate, thereby reducing the psychological distance between present and future selves,” Vlaev added. For the professor, the chatbot embodies the idea of a nudge or subtle intervention designed to guide behaviour in beneficial ways, that could lead users to make better decisions by making the future self more relevant to the present.

Improved wellbeing

In a pre-print paper which hasn’t been peer-reviewed yet, the MIT team tested their application on 344 people aged between 18 and 30. They observed that “users reported decreased anxiety” after having a conversation with their digital older selves as well as lower “negative emotion” or feeling unmotivated. Though further research is needed to know more about the results, Vlaev said that effectiveness will likely depend on the chatbot’s ability to sound authentic. “The use of a digitally-aged image is a clever touch, as visualising oneself in the future can further enhance the sense of continuity and realism,” he said. “However, it is not clear whether people have done anything different, changed their behaviour, as a result of interacting with their future self,” Vlaev noted. Researchers noted in the study’s section on ethical considerations to be mindful of AI-generated content that could endorse negative or harmful behaviours.