Vectors are the new JSON in PostgreSQL

I don't understand why everyone keeps saying that no one is going to want to use PostgreSQL for inference... of course not, no one is suggesting that. But they are going to want to store embeddings of their data and be able to query those embeddings by similarity. Just like they do today in the various existing vector databases.

Years go by, and I, a simple webdev basement dweller, just can’t stop looking at Postgres and think once in a while: “man, what <i>can’t</i> this thing do?”<p>(Of course it can’t do a lot of things, but the stuff that’s baked into it already or can be enabled by some extensions is simply amazing)

I think one of the most important and interesting questions regarding using pgvector is performance, and in particular, how it compares to Qdrant, Weaviate, etc. This post hints at that where it says &quot;I presented a lightning talk called Vectors are the new JSON where I shared use-cases and some upcoming challenges with improving PostgreSQL and pgvector performance for querying vector data. Some problems to tackle (many of which are in progress!) involve adding more parallelism to pgvector, adding support for indexing for vectors with more than 2,000 dimensions, and leverage hardware acceleration where possible to speed up calculations.&quot;. But the post doesn&#x27;t give any numbers. I tried to read the linked lightning talk, but it&#x27;s on <a href="https:&#x2F;&#x2F;www.slideshare.net&#x2F;" rel="nofollow noreferrer">https:&#x2F;&#x2F;www.slideshare.net&#x2F;</a>, which is a really weird website now -- every few slides it tried to force me to watch a 30 second commercial!?

Someone wants to get started in AI&#x2F;ML today and they have beginner-level understanding of Python&#x2F;Javascript. Without any further context, but a desire to learn AI&#x2F;ML and building on what they know should that person next look to: 1) learn PostgreSQL, pgvector, and whenever the &quot;new&quot; comes 2) learn PyTorch, TensorFlow in Python 3) learn TensorFlow.js Presume hobby-level interest, not production-safe best practices - so I guess there is that additional context

Problem is not all managed postgre services support pgvector

what about tensors? do anyone know any pointers on that? Working with tensors instead of vectors