Answers are grounded in this paper and cite their sources.
Artificial intelligence has changed what a biological model can be. The digital twin, a computational replica kept in step with its living original through continuous data assimilation, is now being built at every level of biological organization, from the single cell to the whole population. I argue that deep learning, transformer architectures, physics-informed neural networks, generative models, and the first biological foundation models have moved the twin from a static system of differential equations toward an adaptive, human-relevant instrument, and that this shift is the most promising route yet to replace animal experiments with New Approach Methodologies. I trace five nested scales: the cellular AI Virtual Cell, the in vitro process and organoid twin, the physics-informed organ twin, the individual Virtual Human Twin of precision medicine, and the population twin of the Exposome and virtual control groups. At each level, I ask what the twin genuinely replaces today and what remains aspiration, resisting the temptation to sell a lifetime replica of a human that no one has yet built. The twin will not earn scientific and regulatory trust on cleverness alone. It needs next-generation validation, agreed credibility standards, and lifecycle governance. That, and not more computing, is the work ahead.
No takes yet. Share an insight, caveat, or question.
Thomas Hartung (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: