Deciding whether to change careers, move cities, or care for an aging parent is not simply a matter of gathering more information. Yet that is precisely what most people assume: that hard choices are hard because they lack data. A growing body of research, and a new wave of AI tools designed around it, suggests the reality is considerably more complicated.
The Idea Behind Aging You by 30 Years
Researchers have long observed that people who feel a strong psychological connection to their future selves tend to make better long-term decisions, from financial planning to health behaviors. The problem is that most people experience their future self as a stranger, someone abstract and distant enough to ignore when a tempting short-term option appears.
Low-tech interventions have tried to close that gap for years: writing letters to an older version of oneself, or running vivid imagination exercises. The hypothesis behind AI avatars is that a more concrete, visual, and interactive simulation could do the same job more effectively.
Pat Pataranutaporn, an AI-human interaction expert at MIT, and colleagues are testing exactly that. Their approach uses AI tools to digitally age a person’s image by 30 years, animate it, and then feed the system detailed questionnaires the participant has filled out. The result is an avatar capable of generating hypothetical but plausible future “memories,” powered by Claude Sonnet 4.5. The avatar does not just look like an older version of you. It can hold a conversation grounded in your actual stated values and circumstances.
What the Numbers Showed, and What They Did Not
In a study presented at the 2026 Proceedings of the Augmented Humans International Conference, Pataranutaporn’s team divided nearly 200 participants facing a real personal decision into four groups. Some spoke with one avatar representing a possible future path. Others spoke with two or three avatars, each embodying a different life trajectory. A control group simply imagined their future selves without any avatar interaction.
The most striking result came from the group that spoke with three avatars. Twenty percent of those participants said they would choose an AI-generated option they had not previously considered, compared with less than 3 percent in the control group. That is a substantial gap. It suggests that exposure to multiple simulated futures can genuinely expand the range of options a person is willing to entertain.
The caveats, though, are significant. Whether participants actually followed through on those inclinations after the experiment is unknown. Decision scientist Simon van Baal of the University of Leeds raises a pointed concern: marketing research shows that people can be drawn to options simply because they feel novel, not because they are genuinely better. The AI-generated suggestion may tap into something real, or it may be exploiting a cognitive bias toward the shiny and new. Van Baal also notes that the study’s control group reflected alone rather than speaking with another person, which makes it harder to know whether the avatar is outperforming human conversation or simply outperforming silence.
Social psychologist Anne Wilson of Wilfrid Laurier University adds another layer of caution. AI tools, she notes, are known to “blow a lot of smoke,” meaning they can reinforce fantasies rather than help people pursue genuinely attainable goals. A chatbot that flatters is not the same as a chatbot that helps.
The Deeper Problem These Tools Have Not Solved
Here is what most coverage of AI decision-support tools tends to miss. The assumption embedded in most of these systems is that hard choices are hard because of uncertainty and fear, and that the solution is therefore more information, more simulation, more data. Philosopher Ruth Chang of the University of Oxford argues that this framing is fundamentally wrong.
Writing in the Spring 2017 Journal of the American Philosophical Association, Chang contends that the hardest choices are hard not because one option is clearly better and we cannot see it, but because no option is objectively superior. The alternatives are, in her term, “on a par.” Choosing between caring for an aging parent and pursuing a demanding career is not a data problem. It is a values problem. Research published in April 2025 in the Proceedings of the National Academy of Sciences supports this view, finding that people struggle most with decisions that pit core values against each other.
Chang is building her own AI chatbot, but its design reflects a different philosophy. Rather than simulating futures, her tool runs users through extended exercises, some taking hours, that help them examine whether a choice is genuinely hard, how each path might unfold for better or worse, and whether they are actually ready to commit. The goal is not to generate a novel option but to help a person understand what kind of person they want to be.
There is also a fairness dimension worth noting. Preliminary research published in March, cited by behavioral scientist Adrian Camilleri of the University of Technology Sydney, found that a chatbot advised hypothetical higher-income participants to dream big and prioritize long-term gains, while advising lower-income participants to take fewer risks and focus on short-term stability. The avatar, in other words, projected assumptions about who deserves to think expansively about the future.
In Short
AI avatars that simulate future versions of yourself can meaningfully expand the range of options a person considers when facing a major decision. The evidence for that is real. What the evidence does not yet show is whether those expanded options lead to better outcomes, whether the effect persists beyond the experiment, or whether the technology is doing something qualitatively different from a good conversation with a thoughtful friend. The deeper issue, one that avatar-based tools largely sidestep, is that the hardest life choices are not information problems. They are problems of competing values, and no simulation of a future self resolves the question of who you want to become.
Based on reporting from Science News.