Hacker News: biosubterranean

New comment by biosubterranean in "ChatGPT Images 2.0"

biosubterranean — Tue, 21 Apr 2026 22:58:43 +0000

Oh no.

Quantifying Conversational Reliability of LLMs During Multi-Turn Conversation

biosubterranean — Tue, 24 Feb 2026 20:50:42 +0000

Article URL: https://openreview.net/pdf?id=uFlvUriRLj

Comments URL: https://news.ycombinator.com/item?id=47142804

Points: 1

# Comments: 0

New comment by biosubterranean in "Leveraging LLMs to extract smoking history from clinical notes"

biosubterranean — Fri, 13 Feb 2026 00:43:44 +0000

This article also shows how LLMs are able to perform medical labeling tasks at similar accuracy to human annotators through zero-shot prompting.

Academia tends to lag a year or two behind the times, but I believe this is a valuable and extremely practical way LLMs should be used in practice.

Leveraging LLMs to extract smoking history from clinical notes

biosubterranean — Fri, 13 Feb 2026 00:43:44 +0000

Article URL: https://www.nature.com/articles/s41746-025-02009-y

Comments URL: https://news.ycombinator.com/item?id=46997427

Points: 2

# Comments: 1

New comment by biosubterranean in "Discord just killed anonymity"

biosubterranean — Thu, 12 Feb 2026 23:47:17 +0000

Discord has always felt more personal than just voice comms over a game. The way you can see more and more about what your friends are doing - like what song on spotify or how long you're playing fortnite, and how many days in a row you're playing these games.

I feel like it has always been on this path to capture more and more of your data and personally link it to who you are.

New comment by biosubterranean in "Millions of new materials discovered with deep learning"

biosubterranean — Fri, 01 Dec 2023 16:24:18 +0000

Tldr: Breakthrough in application of machine learning with chemical science to generate novel materials.

Millions of new materials discovered with deep learning

biosubterranean — Fri, 01 Dec 2023 16:24:18 +0000

Article URL: https://www.nature.com/articles/s41586-023-06735-9

Comments URL: https://news.ycombinator.com/item?id=38488506

Points: 1

# Comments: 1

New comment by biosubterranean in "Generative AI Policy for California"

biosubterranean — Tue, 28 Nov 2023 05:06:24 +0000

I'm interested in how the state plans on regulating it, but I do believe there should be some sort of policies in place to set a standard.

Generative AI Policy for California

biosubterranean — Tue, 28 Nov 2023 00:47:47 +0000

Article URL: https://www.gov.ca.gov/2023/11/21/newsom-administration-releases-genai-report/