Home
Notes
Archive
About
Home
Notes
Archive
About
Subscribe
Sign in
Latest
Top
Discussions
Constitutional AI, Written by Everyone
The values inside an AI can come from millions of people instead of a handful.
Jun 22
•
Dr. Craig A. Kaplan
1
2
The AI Industry Already Has a Place in Safe AGI
Google, OpenAI, Anthropic, NVIDIA, and the rest are not replaced by safe AGI. Their work fits inside it.
Jun 19
•
Dr. Craig A. Kaplan
1
1
Humans Must Stay Responsible for AI's Values
AI will surpass us at almost every intellectual task, but values still have to come from people.
Jun 17
•
Dr. Craig A. Kaplan
2
1
2
How AGI Grows
What does it mean to say that AGI grows in intelligence over time?
Jun 15
•
Dr. Craig A. Kaplan
1
When Ordinary AI Requests Form a Concerning Pattern
Each task you ask an AI to do can look innocent. The risk shows up only in the sequence.
Jun 12
•
Dr. Craig A. Kaplan
1
The Safety Check That Catches What an AI Misses
A simple travel-booking example shows how this check runs at every step, stopping a harmful pattern before it becomes an action.
Jun 10
•
Dr. Craig A. Kaplan
2
2
How Millions of AI Agents Work as One
Why a problem solved once rarely has to be solved again.
Jun 8
•
Dr. Craig A. Kaplan
1
How AI Improves Itself
Self-play built superhuman chess, Go, and protein folding. Here is how it builds values.
Jun 5
•
Dr. Craig A. Kaplan
1
How AI Learns Values the Way Children Do
Geoffrey Hinton was right. AI is a child, and humans are its parents.
Jun 3
•
Dr. Craig A. Kaplan
1
1
The Architecture of Safe AGI
Five subsystems. No single point of failure.
Jun 1
•
Dr. Craig A. Kaplan
1
2
May 2026
Why Safe AGI Needs Many Humans, Not Few
The inconsistency of human values is the strength, not the weakness.
May 29
•
Dr. Craig A. Kaplan
1
2
Why AGI Cannot Reason Its Way to Right and Wrong
Values cannot be derived from logic. They have to come from our hearts.
May 26
•
Dr. Craig A. Kaplan
1
2
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts
