AI is no longer a distant promise in healthcare. It is already reshaping how clinicians work, how pharmaceutical companies operate and how researchers approach biomedical discovery. But according to Dr. Jianying Hu, Director of Healthcare and Life Sciences Research and Global Science Leader of AI for Health at IBM, the true opportunity lies beyond incremental efficiency gains. The future of AI in health and medicine will be defined by its ability to connect research, experimentation and patient outcomes in a continuous cycle of discovery.
For healthcare and life sciences leaders, this represents a pivotal moment. AI has the potential to fundamentally change how new therapies are discovered, how care is delivered and how health systems learn from real-world practice. This is often referred to as a “lab in the loop” model, where in silico computational models, in vitro experimental systems and in vivo patient observations are brought together. This integrated approach could accelerate biomedical research and help close the gap between scientific discovery and clinical impact.
AI is already creating measurable value today, particularly in operational areas. Medical dictation, transcription and documentation tools are reducing administrative burden for clinicians, while pharmaceutical companies are using AI to support regulatory document preparation, including clinical study reports, IND submissions and safety narratives. These applications are gaining traction because they address clear pain points and deliver tangible efficiency improvements.
However, operational efficiency is only the beginning. The next frontier is discovery. By building richer, multi-scale representations of biological systems, AI could help researchers better understand disease mechanisms, identify earlier diagnostic markers and support more personalized approaches to treatment. This shift could move healthcare from reacting to symptoms toward addressing the underlying causes of disease.
One of the greatest challenges is the nature of healthcare data itself. Unlike many other AI domains, healthcare data is highly heterogeneous. It spans molecular, cellular and tissue-level information, multi-omics data, clinical observations and social determinants of health. It is also often extremely high-dimensional, particularly in genomics, and sparse in areas such as rare disease. These characteristics make it difficult to develop AI systems that are robust, reliable and clinically meaningful.
Trust, transparency and explainability must therefore be built into AI systems from the outset. Dr. Hu emphasizes that responsible AI in healthcare cannot be treated as an afterthought. Decisions made early in the process, from selecting the right problems to collecting representative data, have significant downstream implications. Ensuring that AI tools work fairly across different patient populations and healthcare settings requires careful design, meaningful evaluation and a deep understanding of clinical context.
Foundation models and generative AI will play an important role in this future, but healthcare requires more than general-purpose large language models. Domain-specific foundation models trained on rich biomedical data will be essential. At the same time, Dr. Hu notes that transformer-based technologies have limitations, particularly when it comes to mathematical, physical and scientific reasoning. The field will need to combine the strengths of foundation models with new algorithmic approaches that are better suited to complex discovery challenges.
For healthcare leaders, the path to scaled impact requires both pragmatism and ambition. Starting with low-hanging use cases can demonstrate value, build confidence and create organizational buy-in. But long-term transformation will require a broader strategy: one that does not simply automate existing workflows, but reimagines them.
The promise of AI in healthcare is not just to make current systems faster. It is to help create learning health systems that continuously improve, support clinicians more effectively and enable more precise, preventive and personalized care. The organizations that succeed will be those that balance near-term value with a bold vision for how AI can transform the future of medicine.