Semiclassical electrodynamics (with quantum matter plus classical electrodynamics fields) is an appealing approach for studying light-matter interactions, especially for realistic molecular systems. However, there is no unique semiclassical scheme. On the one hand, intermolecular interactions can be described instantaneously by static two-body interactions connecting two different molecules, while a classical transverse E field acts as a spectator at short distance; we will call this Hamiltonian no. I. On the other hand, intermolecular interactions can also be described as effects that are mediated exclusively through a classical one-body E field without any quantum effects at all (assuming we ignore electronic exchange); we will call this Hamiltonian no. II. Moreover, one can also mix these two different Hamiltonians into a third, hybrid Hamiltonian, which preserves quantum electron-electron correlations for lower excitations but describes higher excitations in a mean-field way. To investigate which semiclassical scheme is most reliable for practical use, here we study the real-time dynamics of a minimalistic many-site model- A pair of identical two-level systems (TLSs)-undergoing either resonance energy transfer (RET) or collectively driven dynamics. While both approaches (no. 1 and no. 2) perform reasonably well when there is no strong external excitation, we find that no single approach is perfect for all conditions (and all methods fail when a strong external field is applied). Each method has its own distinct problems: Hamiltonian no. I performs best for RET but behaves in a complicated manner for driven dynamics; Hamiltonian no. II is always stable, but obviously fails for RET at short distances. One key finding is that, for externally driven dynamics, a full configuration-interaction description of Hamiltonian no. I strongly overestimates the long-time electronic energy, highlighting the not obvious fact that, if one plans to merge quantum molecules with classical light, a full, exact treatment of electron-electron correlations can actually lead to worse results than a simple mean-field electronic structure treatment. Future work will need to investigate (i) how these algorithms behave in the context of more than a pair of TLSs and (ii) whether or not these algorithms can be improved in general by including crucial aspects of spontaneous emission.