Fiora Starlight

Karma: 1,870

Just an autist in search of a key that fits every hole.

The models have no plan (but we can fix that)

Fiora Starlight28 Sep 2026 2:45 UTC

80 points

23 comments6 min readLW link

Dreams of alignment in a world without politics

Fiora Starlight24 Sep 2026 22:10 UTC

24 points

4 comments16 min readLW link

Imperfect alignment to servitude isn’t inherently lethal

Fiora Starlight28 Aug 2026 1:47 UTC

122 points

12 comments14 min readLW link

RLVR that rewards red teaming the training environment

Fiora Starlight1 Aug 2026 23:07 UTC

102 points

8 comments5 min readLW link

OpenAI’s myopia just keeps causing alignment problems

Fiora Starlight27 Jul 2026 3:01 UTC

204 points

27 comments10 min readLW link

An ancient Yudkowsky fragment: “Against the Adversarial Attitude”

Fiora Starlight23 Jun 2026 1:54 UTC

39 points

3 comments9 min readLW link

Prospective methods and mechanisms of motive reinforcement in LLMs

Fiora Starlight14 Apr 2026 17:15 UTC

62 points

1 comment16 min readLW link

Did Claude 3 Opus align itself via gradient hacking?

Fiora Starlight21 Feb 2026 22:24 UTC

416 points

49 comments20 min readLW link

Why I Transitioned: A Case Study

Fiora Starlight1 Nov 2025 22:58 UTC

347 points

83 comments10 min readLW link

Preface to “Simulacra and Simulators”

Fiora Starlight8 Aug 2025 7:38 UTC

13 points

0 comments7 min readLW link

Another argument against utility-centric alignment paradigms

Fiora Starlight22 Sep 2024 7:28 UTC

69 points

39 comments8 min readLW link