As global demographics shift and disease patterns evolve, fundamental changes are occurring in healthcare demands. On one hand, chronic and complex diseases are accounting for an increasing proportion of cases, while on the other, high-quality medical resources remain concentrated in limited regions and institutions, creating a persistent tension between efficiency and accessibility.
Recently, in a dialogue centered on "Medical AI: From Technological Inevitability to Universal Health Coverage," Deng Qingxu, CEO of Sina Finance, engaged in an in-depth discussion with Kevin Kelly, known as the "Father of Silicon Valley Spirit" and founding editor of Wired magazine, and Wang Shirui, CEO of WeDoctor, about the future of healthcare. During the conversation, Wang Shirui introduced the concept of medical AGI for the first time, expressing confidence in its realization within three years. He emphasized that the significance of medical AGI lies in its potential to dramatically accelerate the pace of clinical medical advancements. For instance, the update cycle for clinical guidelines could be shortened from several years to under three months. Such remarkable progress would help extend healthy life expectancy, making the prospect of living to 120 years a tangible possibility.
Kevin Kelly expressed strong agreement, stating that deploying these "second brains" across every facet of the healthcare system constitutes a genuine revolution. WeDoctor, he noted, is positioned at the epicenter of this transformation. The dialogue explored technical pathways, industry boundaries, and the future landscape of healthcare, consistently returning to a fundamental question: What new possibilities will medical AGI bring to human health and longevity?
With the ability to provide medical advice and diagnose conditions, serious diagnostic AI is reshaping healthcare delivery methods.
As a frontline entrepreneur who entered the medical field as early as 2014, Wang Shirui possesses deep insights into every segment of the healthcare industry. He was the first to propose the "Dual-Speed System" in 2023 and launched MedGPT, China's first large medical model approved for providing disease diagnosis and treatment recommendations. That same year, a West China Hospital clinical trial involving over 2,000 real-case double-blind tests validated the diagnostic consistency between AI and human physicians. The trial demonstrated a 96% diagnostic consistency rate between MedGPT and attending physicians from top-tier hospitals, with a hallucination rate below 0.1%, providing crucial validation for AI's entry into clinical diagnosis. In 2025, the "WeDoctor" app for end-users officially launched, marking the expansion of this technological framework into broader public usage scenarios.
When discussing "WeDoctor," host Deng Qingxu directly posed a key question: What exactly can it do?
Wang Shirui illustrated this with a scenario familiar to the average person: when someone feels unwell and needs to see a doctor, how does the system operate? He explained that within the "WeDoctor" system, initial consultations are led by AI agents. The system conducts multiple rounds of questioning centered on symptoms, medical history, and other relevant information. Based on the collected data, it generates a preliminary diagnosis, which is then reviewed by an expert AI agent for a multi-disciplinary consultation. Following this consultation, the lead physician intervenes to review, revise, and approve the treatment plan, providing final medical recommendations and signing off.
The key to this process lies in human-AI collaboration, with each playing their designated role. The AI handles information integration, preliminary assessment, and workflow coordination, while physicians retain ultimate decision-making authority and responsibility. The technology does not diminish the doctor's role but rather frees them from repetitive tasks, allowing greater focus on core clinical judgment. As Wang Shirui stated, "Many tasks are delegated to the machine. The doctor's approval or additional comments are recorded, revisions are made accordingly, and finally, the doctor must sign and seal the document."
This distinction fundamentally differentiates it from general health assistants. In Wang Shirui's view, the latter primarily operate at the level of health education and information explanation, whereas "WeDoctor" aims to complete the "final step" in the healthcare chain—not merely explaining or offering reassurance, but treating illnesses and saving lives, assuming legal responsibility for the outcomes.
As AI takes over a significant portion of foundational work, physicians' energy and attention can be substantially redirected. Consequently, relatively scarce high-quality medical resources have the potential to be utilized more broadly, serving a larger patient population.
Human-AI collaboration represents the most feasible and effective pathway for AI in medicine.
Beyond specific practices, Kevin Kelly's participation introduced a more forward-looking perspective to the dialogue.
As the founding editor of Wired magazine and author of works such as "Out of Control" and "The Inevitable," Kelly has long focused on the relationship between technological evolution and social structures. In his view, artificial intelligence is not merely a sector-specific variable but a foundational force that will profoundly reshape how society operates. Healthcare stands as one of the most significant arenas for this transformation.
After an onsite visit to WeDoctor's facilities, Kevin Kelly expressed strong approval for their practical approach. "The model WeDoctor is pursuing, I believe, is correct: AI plus human," Kelly explained. He further elaborated that the most viable and effective path in healthcare today is not to replace doctors with AI, but to foster a collaborative relationship between the two. AI can process vast amounts of information and possesses extensive knowledge, while human doctors offer irreplaceable advantages in reasoning, judgment, experience, and assuming responsibility. Only by combining both can the healthcare system achieve a balance between efficiency and reliability.
Within a broader real-world context, the significance of this model may be even more profound.
Kevin Kelly pointed out that, on one hand, many medical needs do not require in-person visits and can be addressed via AI-enabled online consultations, providing substantial convenience for ordinary people. On the other hand, globally, vast populations still lack access to basic medical services. In such contexts, AI-assisted medical services could become the first point of healthcare contact for many individuals. "For them, having an AI doctor is far better than having no doctor at all." It is in this sense that the value of medical AI extends beyond efficiency gains, beginning to address the deeper goal of universal accessibility.
Wang Shirui proposes Medical AGI, aiming to accelerate clinical progress a hundredfold.
As the discussion deepened, it turned to the highly prospective question of how future medicine will evolve.
Kevin Kelly first offered an important distinction. Drawing an analogy with autonomous driving, he noted that such AI systems are essentially "specialized intelligence"—excelling at specific tasks but unable to transfer capabilities freely across different domains like a General AI would. "It's very good at driving, but otherwise quite 'dumb'," he remarked.
Wang Shirui was the first to introduce the concept of "Medical AGI." He suggested that while general AGI might still face significant uncertainty, building systems with similar capabilities within vertical domains—termed "Vertical AGI"—presents a more realistic possibility. This pathway has become a mainstream consensus in current technological development. Demis Hassabis, head of Google DeepMind, has previously indicated that rather than attempting to create a fully general AGI form, breakthroughs are more likely achieved by progressively building system capabilities within specific vertical domains to solve concrete problems.
Historically, medical progress has heavily relied on the accumulation of expert experience, case observation, and clinical trials, with guideline updates typically taking 3-5 years. Medical AGI could enable AI to participate more deeply in patient enrollment, data collection, hypothesis generation, efficacy summarization, and real-world validation, potentially compressing this cycle from years to under three months. At that point, AI would transform not just a single diagnostic step, but the very mode of medical knowledge production. With the assistance of Medical AGI, human understanding of diseases and treatment methods could evolve at a hundred times the current pace. This implies that medical AI can do more than enhance the efficiency of a single consultation; it can fundamentally reshape the evolutionary rhythm of medical knowledge itself.
Kevin Kelly summarized this as the introduction of "second brains"—when AI is embedded into every link of the healthcare system, handling data processing and inference capabilities, medical development will enter a new era. "The medical industry will be utterly transformed by this revolution of comprehensive AI integration, and WeDoctor is right at the center of this revolution."
Conclusion
As the dialogue drew to a close, the focus shifted from technology and trends back to a more fundamental level: What does progress in medicine truly signify?
In Wang Shirui's view, Medical AGI will undoubtedly be realized within three years. This change is gradually turning a long-held vision into reality: high-quality medical resources have the potential to reach a wider population, enabling more people to enjoy longer, higher-quality lives, with greater scope for exploration and choice.
Kevin Kelly shared this sentiment profoundly: "Medical progress doesn't just allow us to live longer; it allows us to live healthier, with better quality, and more happily." In the age of artificial intelligence, this process is being accelerated further, expanding the possibilities of life itself.
As healthcare transitions from passive response to active management, from resource inequality to universal accessibility, and from human-AI collaboration in "treating illness" to a paradigm shift in clinical guideline updates, this AI-driven transformation may just be beginning.
Comments