Struggling with a hard life choice? AI future selves have tips

In each installment of the Back to the Future movie franchise, a time machine drives the same general script: The main character must rewrite a moment in time to fix his family and friends’ lives.

While there’s no real way to see our futures, animations and chatbots are inching us closer to that possibility. “These technologies can actually create a simulation of a time machine by aging the person and helping them see a future version of themselves,” says Pat Pataranutaporn, an AI-human interaction expert at MIT.

Though some may find the tech creepy, many researchers are intrigued. People who can draw a mental line connecting where they are now and where they want to go – known as future self-continuity – tend to act in ways that will benefit their future selves, such as saving more money, considerable research shows. People who view their future selves as strangers, in contrast, often muddle through decisions and may experience intense regret.

Scientists have long sought to help people connect with their future selves, such as through vivid imagination exercises or writing a letter to an older version of oneself.

Relative to those low-tech interventions, researchers suspect that newfangled avatars could help people think more concretely think about their futures, with important caveats. A chatbot that helps people envision attainable goals could help; a chatbot that reinforces people’s unattainable fantasies could do the opposite, says social psychologist Anne Wilson. Problematically, she adds, AI tools are known to “blow a lot of smoke.”  

And, as two distinct AI designs show, avatars project the worldview of their creator, underpinning the disparate ways machines can guide people through their potential futures.

A virtual time machine influences people’s choices

People are turning to AI for advice, with one survey suggesting 1 in 3 U.S. adults has used chatbots for help with a career decision. Yet such advice usually lacks personalization. “The more information any algorithm has about you, the more it can predict what you might value,” says Wilson, of Wilfrid Laurier University in Waterloo, Canada.

To that end, Pataranutaporn and colleagues are using various AI tools to age people’s images by 30 years and then animate them. The team is also feeding Claude Sonnet 4.5 detailed questionnaires filled out by individuals. With that visual and contextual information, avatars can generate hypothetical but realistic future “memories.”

The team’s prior work suggests that chatting with one wizened digital version of themselves helps young people bond with their future selves and envision their lives in a decade.

This time, the researchers divided almost 200 participants facing a tough decision into four groups that interacted with various avatars. The numbers of avatars people chatted with varied. A person pondering career choices might spend a few minutes chatting with an avatar who became a doctor, avatars who became a doctor or an engineer or avatars who became a doctor, engineer or a bioengineering professor, a novel AI-generated suggestion. A fourth group, the control, simply imagined their future selves as they weighed a life decision. Research shows imagining negative futures can increase mental distress, so the team focused solely on positive outcomes.

Strikingly, participants who spoke with the three avatars said they’d choose the AI-generated suggestion 20 percent of the time, the team reported at the 2026 Proceedings of the Augmented Humans International Conference. That’s compared with less than 3 percent of participants in the control condition, who were simply asked if they would select a previously unconsidered path.

Whether people stick with their inclinations after the experiment remains to be seen, Pataranutaporn says. But the AI-generated option “shows how people can really imagine futures which they would not have thought of by themselves.”

Not everyone is as positive. The novel option may tap people’s deeper desires or manipulate them into following a bad path, says decision scientist Simon van Baal of the University of Leeds in England. Marketing research shows that companies can lure people to subpar options because they’re shiny and new, he says. “Whether AI is really adding something here, that’s the question. That’s not really answered by this paper.”

Identifying ingredients of a hard choice might build a bot guide

Other criticisms run deeper – to the root of why people struggle with decisions in the first place.

Most people, social scientists included, assume that choices about the future are hard because they entail uncertainty and fear, says philosopher Ruth Chang, of the University of Oxford in England. The best path forward, then, is to gather as much information as possible, potentially with help from a chatbot.

That framing is wrong, Chang argues in the Spring 2017 Journal of the American Philosophical Association. Fear and uncertainty play a part. But the central issue with a hard choice is that no option is objectively superior, yet any choice sets people on markedly different life paths. The options are “on a par,” Chang says.  

People do really struggle most with decisions that involve choosing between conflicting core values, researchers reported in April 2025 in the Proceedings of the National Academy of Sciences. For instance, should a person care for their aging family or pursue a high-flying career? 

Chang is developing an AI chatbot to help people navigate gut-wrenching decisions. Her tool will run people through exercises that take hours to complete. One exercise pushes people to reflect on whether the decision before them is truly hard. Another runs through how options might play out for the better – or worse.

“My … agent helps you think about a choice,” Chang says. “Can you commit to one kind of path as opposed to another?”

Users may realize they aren’t ready to commit. Then, the rational, if uncomfortable, choice may be to wait, Chang says. For those ready to commit, Chang says, the decision can be empowering, affirming identity with a choice that declares: “I am this kind of person.” 

Avatar or human, that’s the question

Avatars from the future are flashy, but are they better than humans? That’s unclear, says van Baal, who wishes the MIT study’s control group talked with a living, breathing person instead of mulling their thoughts alone. “That’s what people in our lives do all the time,” he says.

But if they’re speaking with like-minded people, they probably aren’t seeing all the possibilities, says behavioral scientist Adrian Camilleri. Preliminary research published in March suggests that “AI tools can act as a sounding board to offer unusual or disconfirming advice that broadens a person’s perspective,” says Camilleri, of the University of Technology Sydney.

Except there’s a hitch. In that provisional work, at least, the chatbot advised hypothetical higher-income participants to dream big and focus on long-term benefits. It advised lower-income participants to take less risks and focus on short-term stability. 

The time machine in the Back to the Future movies is prone to glitching in unforeseen ways, in one instance sending its inventor to 1885 instead of 1985. Is it so surprising, then, that virtual time machines might also emit the occasional burp?