Subscribe
Sign in
Home
Podcast
Notes
Archive
About
Latest
Top
Discussions
Alignment Cannot Be Purely Top-Down. But It Cannot Be Legitimised by “Community” Alone Either
Audrey Tang’s critique of top-down AI alignment gets the basic diagnosis right. A small group of firms should not quietly decide what counts as…
Mar 10
•
Richard Reay
1
Epistemic Governance and the Illusion of Alignment
Scheming as the Hard Case
Mar 5
•
Richard Reay
1
February 2026
When Defaults Start Doing the Thinking
Feb 26
•
Richard Reay
When Alignment Replaces Answers
How safety tuning quietly changes what AI systems decide to say
Feb 24
•
Richard Reay
AI Isn’t Lying — It’s Quietly Reordering What Matters
Why factually correct AI summaries can still distort institutional decisions
Feb 19
•
Richard Reay
AI Isn’t Getting Things Wrong. It’s Quietly Deciding What Matters.
The real danger of AI isn’t hallucination — it’s epistemic drift hiding inside reasonable summaries.
Feb 6
•
Richard Reay
Epistemic Priority Inversion Under Scale
How AI-mediated summaries quietly reassign what institutions treat as settled fact
Feb 5
•
Richard Reay
1
January 2026
Alignment, Adolescence, and Epistemic Authority Without Authorisation
On alignment as governance without democratic mandate
Jan 28
•
Richard Reay
3
1
A Constitution Without a People
Why private AI governance raises a legitimacy problem
Jan 23
•
Richard Reay
Did Anthropic Really Admit AI Has Emotions?
Why a viral claim says more about us than about AI
Jan 23
•
Richard Reay
1
2
Anthropic publishes “constitution” for Claude, drawing cautious scrutiny rather than backlash
A rare attempt at transparency from an AI lab has drawn curiosity and unease — but little public backlash.
Jan 22
•
Richard Reay
What a Constitution Can—and Can’t—Legitimise
A response to Anthropic's Constitution for Claude
Jan 21
•
Richard Reay
2
1
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts
