Post #2173339
2026-04-29 20:11 UTC
@fskornia that is close to my thinking & perhaps why it doesn't feel like there are truly exciting developments. It is hard to distinguish them from AI hype/extraction.
I mention semantic search because it's intriguing to finally be able to query with natural language & not need to enumerate synonyms or know domain-specific terms; the information retreival holy grail. It's its own technology, using some of the same theory as LLMs, but without sharing the ideological commitments of AI companies.
Replies (2)
-
@hugh@ausglam.space 2026-04-29 20:44
@phette23 I think vector search will be the actually-useful legacy of this moment. Anything else exciting (positive) will be political, not technological. Vendor reliance and consolidation is the biggest thing to address.
-
@linguistory@code4lib.social 2026-04-30 20:48
@phette23 @fskornia Random question, but do you know if anyone has looked into how well vector search holds up with historical sources? I’m assuming all the vectorizers are trained on fairly modern texts, which seems like it might cause problems, given how language and concepts drift.