Artificial intelligence, which has become widespread in all fields of science and technology in recent years, has taken its place as an alternative method in the field of nuclear physics. Machines, which are subjected to learning with the use of existing data, can make predictions on what they have learned, and can complete the future data or the deficiencies in the data set it belongs to. From this point of view, it seems possible to carry out a nuclear physics experiment without the need for any experimental setup. For example, by using the binding energy information of about 3000 isotopes whose experimental data are available in the literature, it can be ensured that computers receive a good training on the binding energies of atomic nuclei. This training will open the door for us to correctly request information about any isotope whose experimental binding energy data is not available in the literature. Before artificial intelligence, computers could give us what we gave them. However, with artificial intelligence, we can now take more than we give. Because as we said, computers are learning now. Basically, they perform this learning with the artificial neural networks (ANN) method, which models the operation of the human brain. In this talk, how ANN is used as a nuclear physics laboratory will be discussed with examples such as the determinations of the nuclear ground and excited state energies, nuclear radius, beta decay energies of the isotopes, nuclear reaction cross-sections and single-particle energies for the nuclear structure calculations.