Understanding the three-dimensional structure of the diffuse interstellar medium (ISM) - the density and velocity structure of gas and dust, and the magnetic fields that threads it - is essential for unraveling Galactic structure, star formation, and the Milky Way’s dynamical ecosystem. Modern wide-field surveys now provide distance-resolved dust extinction, spectroscopic absorption, velocity-encoded emission from atomic and molecular gas, Galactic Faraday-rotation, and polarized-starlight absorption measurements, offering an unprecedented view of the local ISM. I will give an overview of recent and ongoing efforts to map these absorbing and emitting components in 3D position and 4D position–velocity space, and describe the machine-learning methods that make such reconstructions possible at scale. In particular, Gaussian Processes offer a natural framework for encoding spatial correlations and prior knowledge about the diffuse ISM, while advanced variational-inference techniques allow us to directly sample from the resulting high-dimensional volumetric posteriors. I will show how these tools enable high-resolution 3D maps of dust, gas, velocity fields, and the Galactic magnetic field, and how combining multiple tracers within a unified probabilistic model yields a coherent, distance-resolved, and uncertainty-aware picture of the Milky Way’s diffuse structure. This approach provides new astrophysical insights today and lays the foundation for the next generation of photometric, spectroscopic, and multi-modal datasets available to us in the near future.
