Employing femtosecond laser pulses [1.59 eV, $(80\ifmmode\pm\else\textpm\fi{}5)\mathrm{fs},$ 80 MHz], we perform time-dependent second-harmonic (SH) measurements to probe the dynamics of charge-carrier transfer processes across native ${\mathrm{S}\mathrm{i}/\mathrm{S}\mathrm{i}\mathrm{O}}_{2}$ interfaces in a new peak intensity regime up to $100{\mathrm{G}\mathrm{W}/\mathrm{c}\mathrm{m}}^{2}.$ The SH signal increases steadily during irradiation up to $45{\mathrm{G}\mathrm{W}/\mathrm{c}\mathrm{m}}^{2}$ laser peak intensity, but changes its temporal evolution above $45{\mathrm{G}\mathrm{W}/\mathrm{c}\mathrm{m}}^{2}:$ as a different phenomenon, the SH signal surpasses a maximum, decreases slowly, and appears to approach a steady state level. We assign this observed signal decrease to the light induced injection of holes into the ${\mathrm{SiO}}_{2}$ thin layers involving a four-photon process, which opposes the electron contributions (three-photon process). Furthermore, we investigate the SH signal development after irradiation interrupts and find that preirradiated samples show a drastically accelerated SH response. Both, the electron and the hole injection processes appear to be about two orders of magnitude faster in preirradiated than in ``virgin'' samples. We assign this acceleration to the laser induced generation of permanent defects, which affect electron as well as hole trapping in the ${\mathrm{SiO}}_{2}$ layers. In addition the SH signal amplitude is shown to be a function of the dark period duration, and the typical time constant for the depopulation of hole trap sites under dark conditions has been extracted and amounts to $(110\ifmmode\pm\else\textpm\fi{}15)\mathrm{s}.$ We present an empirical model involving four exponential functions, taking into account the electron and hole contributions to the time-dependent SH response. This model fully reproduces the recorded experimental data. It supports our interpretation of opposing three- and four-photon processes being involved in electron and hole injection, respectively. Finally, we employ SH imaging in comparison with scanning electron microscopy to visualize sample areas with laser defects.