${}^6\text{Li}$ is known experimentally (Pyykkö 1991) to have a very small quadrupole moment, $-0.082\ \text{fm}^2$. This indicates a spherical shape. So as a start we could try just using the spherical single-particle levels, ignoring residual interactions, and filling levels up to the Fermi level for both protons and neutrons. Doing this, we find that the configuration of the valence particles would be $\pi p_{3/2}\otimes \nu p_{3/2}$. This would produce a degenerate multiplet of states with spin-parity values $0^+$, $1^+$, $2^+$, and $3^+$. We expect the degeneracy to be lifted by the residual interaction.
In light nuclei, the valence protons and neutrons occupy some of the same states, so neutron-proton interactions can be quite strong. This interaction is a particularly strong attraction when the intrinsic spins of the proton and neutron are coupled to spin 1; this is the reason that the deuteron has spin 1. The spatial overlap between the two $p_{3/2}$ states is maximized if their orbital angular momenta are coupled to spin 0, and this would be expected to maximize the attractive interaction. The 0 orbital angular momentum and the spin 1 from the intrinsic spins gives a total angular momentum of 1. So this would explain the reason why the $1^+$ state is so much lower in energy than the others, and is in fact the only particle-stable state in the nucleus -- the other members of the multiplet are resonances.
A couple of other answers (1, 2) talk about ${}^6\text{Li}$ as a cluster consisting of an alpha and a deuteron. The trouble with these clustering descriptions is that they always violate the Pauli exclusion principle, so I would prefer an explanation in terms of the vanilla shell model, which seems to work here.
Pyykkö, "The Nuclear Quadrupole Moments of the 20 First Elements: High-Precision Calculations on Atoms and Small Molecules," 1991 - https://www.degruyter.com/view/j/zna.1992.47.issue-1-2/zna-1992-1-233/zna-1992-1-233.xml (open access)