feat: Update link log

site/link-log.yaml

@@ -1,3 +1,27 @@
+- url: https://darthmall.net/weblog/2023/rss/
+  name: "Evan Sheehan: RSS?"
+  summary: >
+    There's a swing back to RSS right now, which I think is good. But I also think that Evan's
+    thoughts here are good. RSS can't be the only solution for how we take in the web. It can't
+    be the only solution for how we decentralize the web again.
+
+    Hat tip to [Greg Morris](https://gregmorris.co.uk/2024/01/29/visit-more-blogs) for his related
+    post that helped me find this one.
+- url: https://ethanmarcotte.com/wrote/generative/
+  name: "Ethan Marcotte: Generative"
+  summary: >
+    A collection of quotes about AI from 1683 to 2024.
+- url: https://seldo.com/posts/ai-ml-llms-and-the-future-of-software
+  name: "Laurie Voss: On AI, ML, LLMs and the future of software"
+  summary: >
+    A level-headed explanation of what exactly AI, ML, and LLMs are. "LLMs are really complex markov
+    chains; but the really complex part makes them qualitatively different" has been by go-to
+    explanation of LLMs. I disagree, however, with the idea that LLMs "understand" anything. LLMs
+    contain big statistical models of sentence and paragraph structure, and of the relationship
+    between words and phrases. This allows them to generate text that is more of a statistical match
+    for text written by humans. That humans see this as "understanding" is a form of pareidolia.
+
+    This distinction is narrow, but important.
 - url: https://coryd.dev/posts/2024/towards-a-quieter-friendlier-web/
   name: "Cory Dransfeldt: Towards a quieter, friendlier web"
   summary: >