The issue of "aging alone" has evolved from a social drama into a critical shortage of legitimate solutions. As traditional family ties fray, more patients find themselves in intensive care without an inner circle capable of making life-or-death decisions. The current system of Advance Care Planning (ACP)—static paper declarations of intent—is effectively dead. According to researchers from the Singapore University of Technology and Design (SUTD), these documents sit in archives and remain inaccessible at the exact moment a doctor needs to make a split-second decision. Consequently, life-altering choices are often made by strangers who have never met the patient.
A team led by Assistant Professor Kenny Choo has proposed a pragmatic, albeit ethically sensitive, solution: ACPAgent. This prototype of an autonomous proxy agent doesn’t just store data; it interprets a person’s will dynamically. In a study presented at the CHI 2026 conference, 15 participants trained agents on their core values and risk tolerances. The AI was then tested against scenarios ranging from treatable infections to terminal illnesses. SUTD’s results show an 86.7% agreement rate between the users' preferences and the algorithm's recommendations. This is more than just a "smart notepad"; it is an attempt to create agency where there was previously a vacuum.
Key Study Findings
The AI’s decision-making alignment reached 86.7%, surpassing the consistency of many human proxies.
The agent’s dynamic nature allows it to account for changes in health status, unlike static paper wills.
The technology addresses the lack of legal representatives for the solitary elderly population.
"We are seeing a dangerous drift from reflecting a user's will to constructing it: by helping to find the right words, the agent begins to subtly format the user's values themselves." — Kenny Choo, SUTD.
However, a trap remains: researchers identified a risk of "automation bias." Kenny Choo rightly warns against blind trust in high metrics, pointing out that AI can impose its own structure on responses. Nevertheless, for the insurance and medical markets, this signals a tectonic shift: the legalization of agent-based agency is becoming the only alternative to institutional chaos.
We are entering an era where an AI advocate is becoming an operational necessity. As classic healthcare planning collapses, the choice is no longer between "human and machine," but between informed algorithmic representation and total uncertainty on the operating table. For the business sector, this means the emergence of a new niche for legally significant digital twins whose decisions will carry weight in both the courtroom and the clinic.