The main benefit of a vector database, however, is that it can index the data by proximity, speeding up clustering, partitioning, or nearest neighbor kinds of queries.

"for one _dimensionality reduction_ technique" i should say

Hmm, probably? I haven't done a whole lotta work in this space recently. But there's inherent difficulty in visualizing vectors in very high dimensional spaces (100s or 1000s, ofttimes). Check out "principal component analysis" for one technique