Skip to content
Sonic AI
MO

Musings on the Alignment Problem

Summary

Musings on the Alignment Problem covering AI Safety, AI Alignment, Large Language Models (LLMs), and Automated Alignment Research. Notable guests include Jan Leike. Episodes span from Sep 2022 to Jan 2026.

5episodes
80total claims
12topics covered
5 episodes
Alignment is not solved (but increasingly looks solvable)
Jan 22, 2026

Significant progress was made in aligning large language models during 2025, with models like Anthropic's Opus 4.5 and OpenAI's GPT-5.2 showing marked improvements over predecessors. Simple, target...

AI AlignmentLarge Language Models (LLMs)SuperalignmentReinforcement Learning (RL)+11 more
Should we control AI instead of aligning it?
Jan 24, 2025

The author contrasts two AI safety strategies: 'AI control' (managing a potentially misaligned AI's behavior externally) and 'AI alignment' (building an inherently trustworthy AI). While control te...

AI SafetyAI AlignmentAI ControlMisalignment Risk+11 more
A proposal for importing society's values
Mar 9, 2023

The current method of AI value alignment, where tech companies make unilateral decisions, is unsustainable and risks being driven by commercial incentives rather than societal well-being. The autho...

AI AlignmentLarge Language Models (LLMs)Value AlignmentDeliberative Democracy+11 more
Why I'm optimistic about our alignment approach
Dec 5, 2022
What could a solution to the alignment problem look like?
Sep 27, 2022

Sign up free to see the full analysis

Get started free

Top Topics

Musings on the Alignment Problem, Sonic AI