AI model DenseAV learns through watching videos

AI model DenseAV learns by comparing pairs of audio and visual signals and determines what data is important. To learn these patterns it uses audio-video contrastive learning to associate a particular sound with the observable world. This mode of learning means the visual side of the model can’t gain any insights from the audio side (and vice-versa) forcing the algorithm to recognize objects in a meaningful way. Will AI become smart the geniuses of this world?