AI and health ·

Three Friends at Dinner: Who Answers When Some of Us Leave Our Bodies?

The third in my reflections since ICAIH 2026, after AI Is Not a Tool and Beyond the Faster Horse. First published on LinkedIn.

Illustration: three friends at dinner on a Beirut rooftop at night, one of them in a robot body
AI generated using Gemini.

In my reflections since ICAIH, I've argued that AI is not a tool, that trust is a method, not a feeling, and that duty of care now includes the AI we use. This one takes those ideas into health: care, rights and responsibility when some of us change our bodies, or leave them.

It starts with a story. The scenes are fiction. The facts after them are not.

Beirut

Three friends who met at university sit down for dinner. Christine is 60 and looks it. Her son in Dubai covers the rent, her nurse is a person, and her doctor signs off on diagnoses a machine makes.

John is 60 and looks 35. A one-time gene edit keeps his blood fats low, a weekly drug slows his biological clock, and an implant reports his health to his insurer every hour. His premium falls when he sleeps well and rises when he doesn't. He never signed a form. He tapped "agree" in an app.

Sophie arrives last, in a body she rented for the evening. After an illness at 51, her mind was copied, and it now runs in a data center in the Gulf. Her husband still wears his ring. No one is sure whether the marriage stands, whether she owns her apartment, or whether her health plan covers a mind without a body. When the data center goes down, who is responsible for her?

What's real today

The three lives sit on three horizons: Christine's is arriving now, John's rests on predictions for the next decade, and Sophie's, on expert estimates, is a century or more away. Institutions have to plan for all three at once. Christine's world is the closest. On November 10, Abu Dhabi's Department of Health added an AI physician assistant to Malaffi, the emirate's health records exchange, for all physicians, with medical decisions left to the doctor.

John's world is where the startups are, and where aging is treated as something to fix. Bryan Johnson, the tech entrepreneur who spends about $2 million a year on his regimen, claims he has already reversed his biological age and aims for immortality by 2039. NewLimit is valued at $3.1 billion. On October 8, Life Biosciences reported first human data from a safety trial of partial epigenetic reprogramming, which aims to restore old cells to a younger state: in three glaucoma patients it was well tolerated, and two showed preliminary signs of improvement on vision tests. Loyal is chasing the first FDA approval of a lifespan drug, in dogs; the FDA has accepted its safety section. AI is speeding the search: DeepMind's AlphaGenome 2, released November 11, predicts how DNA variants affect hundreds of human cell types, and a new preprint used machine learning to predict lifespan-extending compounds in worms. When aging clocks advanced more slowly on a GLP-1 drug in a trial, Lilly said that was mostly weight loss, and a new preprint finds one blood-based aging clock may track disease, not aging, in cancer patients. The rest is prediction. In April 2025, Demis Hassabis told 60 Minutes the end of disease may come "within the next decade or so." Ray Kurzweil, who has written about this longer than almost anyone, predicts longevity escape velocity between 2029 and 2035 for people diligent with their health, and a merger with AI by 2045. Peter Diamandis puts it in the 2030 to 2035 timeframe.

Sophie's world is the furthest: no one can copy or transfer a human mind today. Kurzweil's merger and Michio Kaku's The Future of the Mind, which looks toward uploading the brain to a computer neuron for neuron, both point here. So far, researchers have mapped one fruit fly's brain, 139,255 neurons. A new arXiv study of fly brain-wiring models found that some claims hold up while others depend on how modelers tuned the models. Surveyed neuroscientists put their median estimate for emulating a human brain at around 2125.

In the clinic

Medicine has always known who the patient is. That certainty is starting to move: one day a mind may outlive its body. In between, researchers are measuring what drugs like John's do to biological age. Who is the patient, what counts as alive, and who consents, for whom? Sophie consented once, before the copy existed. Who consents for her now? Khalil N Abi Nader, MD MRCOG and Marie Noelle Kallas Chemaly, how would you answer that in your practice?

In the law

The law recognizes people, and legal entities such as companies. My read is that under today's rules, a running copy of a mind would be treated as software, owned by whoever runs it. Ohio has a bill to ban any legal personhood for AI; Argentina has proposed companies run by AI agents, as I wrote in Beyond the Faster Horse; and on October 30, Pope Leo XIV warned that personhood for AI systems "risks obscuring human responsibility." All three concern AI built by people. Sophie is a human mind with no legal home, and she remembers her wedding. Does her marriage stand? Who inherits? Rany Sader, under Lebanese law today, what would Sophie be?

In the boardroom

Every hospital, insurer and device maker serving John and Sophie will need a board that can answer three questions. Who answers when the AI in our care gets it wrong? What data leaves the patient's body, and who owns it? Who asked the patient first? A Tebra survey reported by Healthcare IT Today found 91% of patients want their permission required every time AI is used to support diagnosis or treatment. In my first reflection, on privacy, I said the next frontier is our thoughts. John's implant makes that literal. Ronnie Hammad, which of these three should a board ask first?

Every level leads back to the same rule: decide who answers before deciding what it is. Duty of care now includes the AI that delivers the care, and one day it may include the mind that outlives the body.

Which of the three is your institution ready to serve?

This is my third reflection since ICAIH, after AI Is Not a Tool and Beyond the Faster Horse. Its format, a story first and then the technology behind it, follows Kai-Fu Lee and Chen Qiufan's AI 2041.

Michael Joseph is an AI expert and consultant with operations across the United States and EMEA. He is Founder and Chief Strategy Officer of Epirroi and AI Advisor to the President of the Holy Spirit University of Kaslik (USEK).

Sources

Is your institution ready for what longevity and AI medicine will ask of it?

Start a conversationStrategy, governance, and foresight Executive AI trainingFor leaders, boards and faculty

← More insights · AI Is Not a Tool · Beyond the Faster Horse