Cynddl | Dark Hacker News

user:	Cynddl
created:	January 22, 2013
karma:	1.3k

1.	Our evaluation of OpenAI's GPT-5.5 cyber capabilities(aisi.gov.uk) 2 points by Cynddl 13 days ago \| 0 comments
2.	Making AI chatbots friendly leads to mistakes and support of conspiracy theories(theguardian.com) 93 points by Cynddl 18 days ago \| 80 comments
3.	UK Biobank health data keeps ending up on GitHub(biobank.rocher.lc) 197 points by Cynddl 25 days ago \| 57 comments
4.	2 points by Cynddl 34 days ago \| discuss
5.	ChatGPT Edu feature reveals researchers' project metadata across universities(fastcompany.com) 2 points by Cynddl 66 days ago \| 0 comments
6.	AI no better than other methods for patients seeking medical advice, study shows(reuters.com) 3 points by Cynddl 97 days ago \| 0 comments
7.	AI chatbots pose 'dangerous' risk when giving medical advice, study suggests(bbc.co.uk) 4 points by Cynddl 97 days ago \| 2 comments
8.	Show HN: Small, anonymous app for teams to do retrospective sessions(retrospective.rocher.lc) 1 points by Cynddl 106 days ago \| 0 comments
9.	Measuring What Matters: Construct Validity in Large Language Model Benchmarks(arxiv.org) 1 points by Cynddl 188 days ago \| 0 comments
10.	AI Capabilities May Be Overhyped on Bogus Benchmarks, Study Finds(gizmodo.com) 43 points by Cynddl 191 days ago \| 17 comments
11.	AI's capabilities may be exaggerated by flawed tests, according to new study(nbcnews.com) 3 points by Cynddl 192 days ago \| 0 comments
12.	Experts find flaws in tests that check AI safety and effectiveness(theguardian.com) 3 points by Cynddl 195 days ago \| 0 comments
13.	Measuring What Matters: Construct Validity in Large Language Model Benchmarks(oxrml.com) 3 points by Cynddl 195 days ago \| 2 comments