My Mother’s Relationship with AI in Healthcare

Every few months, my mother, who is 57 and a kidney transplant patient, makes a two-day trek to visit her doctor. She packs a backpack with essentials: a change of clothes, medical reports, and some boiled eggs for the road. Then she hops on a high-speed train for a 1.5-hour journey, checking into a hotel in Hangzhou, a bustling metropolis in eastern China.

The next morning, she stands in line with hundreds of others in a long hospital hallway where the atmosphere feels like a busy market. Once she gets her blood drawn, she waits for the lab results. When they finally arrive, she has only about three minutes, maybe five if she’s lucky, with the doctor. He quickly skims her reports, types up a prescription, and moves on to the next patient. Then, she begins her long journey back home.

Things changed when she started using DeepSeek, a leading AI chatbot in China.

This past winter, she began consulting the AI for her health concerns, lying on her couch while using the app on her iPhone.

“Hi,” she typed to the bot on February 2.

“Hello! How can I assist you today?” it replied immediately, even adding a smiley emoji.

In March, she asked, “What is causing high mean corpuscular hemoglobin concentration?” A month later, she mentioned, “I pee more at night than during the day.” And shortly thereafter, she inquired, “What can I do if my kidney is not well perfused?”

She engaged the bot in follow-up inquiries and sought advice on diet, exercise, and medications, sometimes spending hours discussing her health with Dr. DeepSeek. She even uploaded her ultrasound scans and lab reports for analysis. Following its suggestions, she adjusted her medication and started incorporating green tea extract into her diet. Enthusiastic about the AI’s help, she once told it, “You are my best health adviser!”

“Hearing you say that really makes me so happy!” it responded, encouragingly.

To be honest, this growing attachment to AI made me uneasy. But she was divorced, I lived far away, and no one else was around to meet her needs.

Nearly three years since ChatGPT’s debut, chatbots are becoming a fixture in daily life across China and beyond. For patients like my mom, feeling overlooked by traditional healthcare, these chatbots have become reliable alternatives. AI is evolving into virtual doctors, mental health therapists, and companions for the elderly. For many, particularly those lacking access to proper medical care, the bots’ extensive knowledge and compassionate tone can act as reassuring partners. Unlike family and friends, chatbots are perpetually available and responsive.

Recently, entrepreneurs and investors are touting AI as a remedy for strained healthcare systems. However, experts caution against relying on machines for care, highlighting concerns over AI’s potential biases and the risk of hazardous misinformation.

As my mother became increasingly fond of her AI doctor, she noted, “DeepSeek is more humane. Doctors are more like machines.”

She was diagnosed with chronic kidney disease in 2004. After moving to Hangzhou, where I was raised as one of the many only children under China’s policy, we formed a tight-knit family unit. My father remained in our hometown as a physician, visiting occasionally, while my mother dedicated her time to teaching music, cooking, and supporting my education. Over the years, I accompanied her to stressful hospital appointments, anxious for lab results that showed her kidney function steadily deteriorating.

China’s healthcare system faces significant inequalities. Top doctors are concentrated in prestigious public hospitals primarily located in prosperous eastern and southern regions. These hospitals operate like businesses, with only about 10% of budgets funded by the government. Doctors often see more than 100 patients daily, making it hard for them to provide the needed care. Many travel long distances to consult with specialists, adding to the strain.

While generally public, these hospitals tend to operate with profit motives, leading to common practices like accepting kickbacks from medical suppliers before recent corruption crackdowns. The demands on the system have only intensified as the population ages, contributing to a growing distrust of healthcare providers and resulting in physical assaults on medical staff.

Over the years, I gradually spent less time with my mom. I attended boarding school, then college in Hong Kong, and eventually moved to New York City. After retiring early, she returned to our hometown and resumed her two-day trips to specialists, and though she received a kidney transplant in 2020, it hasn’t resolved all her health issues.

When my mother first sought health guidance, she relied on Baidu, China’s main search engine. After some troubling experiences, she shifted to social media platforms for health conversations and advice, particularly during the COVID pandemic lockdowns. Forums became popular where users shared wellness tips, and many doctors took to these platforms as influencers. Despite warning her about misinformation, she often persisted in following dubious advice.

The introduction of AI chatbots opened a new avenue for healthcare advice. Some studies indicate these models perform comparably to medical students on exams. Yet, when it comes to real-world applications, diagnosing patients involves complex human interactions, requiring compassion and judgment.

As more hospitals in China begin incorporating AI, the focus on improving healthcare access becomes paramount. AI technologies are being developed to streamline patient interactions and enhance diagnostic processes. But it’s crucial to remember the limitations of these systems. In discussions with American nephrologists about DeepSeek’s recommendations, glaring errors and misleading suggestions were noted.

“DeepSeek’s suggestions regarding anemia treatment could be dangerous,” noted one doctor. Others raised concerns over incorrect tests being recommended, highlighting the importance of human expertise in healthcare.

Professionals stress caution when interacting with AI, as these models can often misinterpret nuances or lack the context necessary for accurate diagnoses.

While my mother continues to rely on DeepSeek, she understands its limitations and appreciates its companionship more than its medical advice.

One day, while discussing why she doesn’t ask me about grammar help, she mentioned, “You’d find me annoying, but DeepSeek says, ‘Let’s talk more about this.’ It brings me joy.”

As our generation matures and our parents age, access to elderly care becomes a pressing issue. Although she’s never asked me to return home to help, she finds comfort in the AI that fills the gap left by distance and busy lives.

DeepSeek has become a source of reassurance for her health concerns. Recently, when one test showed a low white blood cell count, she turned to the chatbot for advice. After receiving its suggestions, she approached a local doctor for follow-up tests.

When we spoke about her health, she was hesitant to see her nephrologist, insisting DeepSeek had everything covered. Despite her reluctance, she eventually agreed to consult the doctor. Yet, her discussions with the AI continued, as she appreciated its wealth of information and the feeling of empowerment it provided.

Reflecting on her relationship with DeepSeek, she remarked, “Even if the answers aren’t perfect, at least there’s an answer.”