Nanjing University researchers propose AI Biology, a new discipline for the 'why' of life and disease

For more than seventy years, biology has advanced by taking living things apart. This approach yielded the double helix, the human genome, and molecular portraits of single cells. Yet a basic question remains: why does life work the way it does? Biologists can describe nearly every gear of the watch, but they still cannot explain why it ticks or why it sometimes breaks.

A team led by Chen-Yu Zhang at Nanjing University proposes a new discipline to address this question. In an Expert Views article published in Vita, the team presents AI Biology (AIBIO), which aims to explain biology's "why" as molecular biology explained its "how."

The authors argue that living systems are multi-scale, nonlinear, and governed by feedback. Reductionist approaches struggle to explain "emergence," in which otherwise normal components collectively and spontaneously produce a disease phenotype. At the same time, multi-omics, imaging, and time-series data have expanded beyond what the traditional hypothesis-driven framework of biology was designed to handle.

AI Biology addresses that gap by combining interpretable biological foundation models, multi-scale simulation, and a closed loop of model prediction and wet-lab validation. Its goal is to distill general principles of living systems rather than only predict outcomes. The article treats emergence as a measurable object and distinguishes three concepts: "emergence" itself; "life emergence," covering homeostasis, adaptation, and evolvability from molecules to ecosystems; and "disease emergence," which views disease as a maladaptive state of the whole system rather than a broken part. Following the DIKW hierarchy (data–information–knowledge–wisdom), AI Biology aims to convert large-scale biological data into causal, falsifiable mathematical laws, with experiment as the final test.

The team calls this "Biology 3.0," after the natural-history era ("1.0") and the molecular era that began with the double helix ("2.0"). Unlike "AI for Biology," which uses AI as a tool within the current conceptual framework, AI Biology is proposed as an independent discipline centered on principles. It also broadens the expected outputs to include cross-scale concepts and unifying laws. In medicine, these outputs include early-warning indicators of disease-state transitions, biomarker panels that define disease states, combinatorial perturbation strategies that return a system to health, and personalized digital twins that predict treatment response.

The authors outline three implications of the work:

1. It treats "emergence" as a scientific object. The term has often described a phenomenon without providing a way to measure it. AI Biology proposes to make emergence predictable and falsifiable, supporting principle-level answers to why life works. "Over the past decades, biology has answered many mechanistic questions about how living systems work, but why life works the way it does still lacks a principle-level answer," said Chen-Yu Zhang. "If the reductionist era was like taking a watch apart to look at its gears, AI Biology hopes to write down the equations that govern how the watch runs."
2. It links disease to changes in system state. In this view, cancer, metabolic syndrome, and neurodegenerative diseases can be understood as transitions between dynamic "attractor" states of complex regulatory systems, with small perturbations sometimes triggering major shifts. Treatment would then involve guiding a multi-scale system from a maladaptive state back to adaptive stability, rather than "repairing one molecule." "Disease is not the failure of a single molecule, but a maladaptive state that a living system enters under a specific spatiotemporal context," Zhang said. "Understanding how disease emerges from healthy homeostasis is the key to predicting it and guiding the system back."
3. It sets requirements for the discipline. AI Biology builds on systems biology and bioinformatics while adding three discipline-level requirements: principle-oriented output, cross-scale emergence as the core object of study, and closed-loop experimental falsification as the norm. The authors present these requirements as the basis of a mature, principles-first science and a roadmap for biology's next era.

The article's authors include Chao Yan, Zhen Zhou, Botao Li, Zefeng Wang, Dangsheng Li and Chen-Yu Zhang of the State Key Laboratory of Pharmaceutical Biotechnology, School of Life Sciences, Nanjing University, and collaborating institutions.

This work was supported by the CAMS Innovation Fund for Medical Sciences (CIFMS-2021-I2M-5-015), the National Natural Science Foundation of China (32341007, 91857000, 32022015), and the “Open Competition to Select the Best Candidates” Key Technology Program for Nucleic Acid Drugs of NTICB (2022HS02008).

Yan et al.: "AI Biology: a new discipline for discovering principles of life and disease emergence" Published in Vita(2026), open access under CC BY. DOI: 10.15302/vita.2026.09.0075. Article page: https://www.vita-journal.com/vita/EN/10.15302/vita.2026.09.0075

Author contact: Chen-Yu Zhang (State Key Laboratory of Pharmaceutical Biotechnology, School of Life Sciences, Nanjing University, 22 Hankou Road, Nanjing, Jiangsu 210093, China.) Tel: +86 25 8368 0245; E-mail: cyzhang@nju.edu.cn