Scattering amplitudes of elementary particles exhibit a fascinating simplicity, which is entirely obscured in textbook Feynman-diagram computations. While these quantities find their primary application to collider physics, describing the dynamics of the tiniest particles in the universe, they also characterise the interactions among some of its heaviest objects, such as black holes. Violent collisions among black holes occur where tremendous amounts of energy are emitted, in the form of gravitational waves. 100 years after having been predicted by Einstein, their extraordinary direct detection in 2015 opened a fascinating window of observation of our universe at extreme energies never probed before, and it is now crucial to develop novel efficient methods for highly needed high-precision predictions. Thanks to their inherent simplicity, amplitudes are ideally suited to this task. I will begin by reviewing the computation of a very familiar quantity Newton's potential, from scattering amplitudes and unitarity. I will then explain how to compute directly observable quantities such as the scattering angle for light or for gravitons passing by a heavy mass such as a black hole. These computations are further simplified thanks to a remarkable, yet still mysterious connection between scattering amplitudes of gluons (in Yang-Mills theory) and those of gravitons (in Einstein's General relativity), known as the "double copy", whereby the latter amplitudes can be expressed, schematically, as sums of squares of the former -- a property that cannot be possibly guessed by simply staring at the Lagrangians of the two theories. I will conclude by discussing the prospects of performing computations in Einstein gravity to higher orders in Newton's constant using a new, gauge-invariant version of the double copy, and as an example I will briefly discuss the computation of the scattering angle for classical black hole scattering to third post-Minkowskian order (or O(G^3) in Newton's constant G).