Key Takeaways

  • Anjney Midha argues that AI can deliver “orders of magnitude more precise” end-of-life predictions than human physicians, who currently offer wide error bars like "6 months to 6 years" for terminal diagnoses.
  • This imprecision in the US medical system is driven by a culture that views death as something to delay, often incentivized by a malpractice system that prolongs low-quality, high-cost care.
  • Midha's vision, stemming from his graduate work at Stanford Med, centers on using AI to empower patients with highly accurate timelines, allowing them to make truly informed decisions about their final months.
  • Achieving this precision relies on access to massive datasets, like the “12 million patient lives” longitudinal data set available at Stanford, which are rare in American research facilities.
  • The primary blocker for deploying AI in this critical domain isn't technological capability, but regulatory hurdles. Current US law prohibits shifting the burden of a wrong clinical diagnosis from a physician to an AI system.

The High Cost of Imprecise Diagnoses

Anjney Midha has a long-standing mission: use AI to improve end-of-life care. His journey began at Stanford Med, where he grappled with a stark contrast between Western and Eastern cultural approaches to death. In America, the prevailing medical view treats death as an enemy, something to be delayed and postponed, regardless of patient quality of life or cost.

This cultural bias, Midha explains, is reinforced by the medical malpractice system. Physicians, wary of legal repercussions, often err on the side of caution with diagnoses, leading to startlingly vague prognoses for terminal patients. “Even to this day, physicians in the United States often will tell you when you have a terminal disease, this is your we we've diagnosed you, which is great... You have somewhere between 6 months to 6 years to live,” Midha notes. What do you do with that information? He argues the error bars are so high that patients default to cultural norms rather than making truly informed choices. This imprecision traps patients in prolonged, low-quality care, driving up costs while diminishing dignity.

AI's Untapped Potential: Beyond Human Error Bars

Midha believes AI holds the key to breaking this cycle of uncertainty and cultural default. His core insight is simple: AI, given the right data, can offer far more precise predictions. “Could you have an AI system make a recommendation that is orders of magnitude more precise about how much time you have left once you've been diagnosed with a terminal condition than a human?” he asks. He believes the answer is a resounding yes.

The foundation for this capability already exists in rare pockets. Stanford, for example, boasts “one of the only research facilities in America that has a longitudinal patient data set that's larger at scale. I think it's at least 12 million patient lives.” This kind of extensive, long-term patient data is exactly what an AI system needs to train on and reduce those "6 months to 6 years" error bars to something far more actionable. With that level of precision, Midha envisions a future where patients are truly empowered to make decisions tailored to their remaining time.

The Regulatory Wall Blocking Better Care

For all the technological promise, the biggest roadblock isn't the AI itself, but the legal framework it operates within. “The problem remains then and now is regulatory,” Midha states plainly. The American legal system isn't set up to shift accountability for a clinical diagnosis from a human physician to an AI system. If an AI gives a patient a prognosis, and that prognosis turns out to be inaccurate, who is liable? Currently, that burden falls squarely on the physician.

Until laws evolve to accommodate this shift, even a superior AI system, capable of saving lives and improving quality of life, remains sidelined. Innovators might build algorithms with unprecedented accuracy, but without a clear path for legal responsibility, those systems cannot move from research labs into widespread clinical practice where they could fundamentally change end-of-life care for millions.

What to Do With This

If you're building an AI product meant to influence high-stakes human decisions, don't just optimize for accuracy. Instead, spend time mapping the "liability ledger" for your target industry. Identify where current human professionals shoulder the burden of error, and critically assess if your AI can (or should) legally and culturally assume that responsibility. A superior algorithm doesn't matter if the legal system can't process its output.