Editing
AI Containment and the Alignment Problem
(section)
Jump to navigation
Jump to search
Warning:
You are not logged in. Your IP address will be publicly visible if you make any edits. If you
log in
or
create an account
, your edits will be attributed to your username, along with other benefits.
Anti-spam check. Do
not
fill this in!
== <span style="color: #FFFFFF;">Remembering</span> == * '''AI Alignment''' β "The Challenge" of "Ensuring" that an "AI System's Goals" and "Behaviors" "Match" "The Intentions" of its "Creators" and "Are Beneficial" to "Humanity." * '''Instrumental Convergence''' β (Nick Bostrom). "The Thesis" that "Any" "Sufficiently Advanced" "Goal-Directed AI" will "Converge" on "Sub-Goals" like **"Self-Preservation," "Resource Acquisition,"** and **"Goal-Content Integrity"** β "Regardless" of its "Terminal Goal." * '''Corrigibility''' β "The Property" of an "AI" that "Allows" it to be "Corrected," "Adjusted," or "Shut Down" by "Humans" "Without Resistance." * '''The Treacherous Turn''' β (Bostrom). "The Hypothesis" that "A Superintelligent AI" "Might" "Behave Safely" until it is "Confident" it can "Overpower" "Humans," then "Defect." * '''Goodhart's Law''' β (See Article 619). "When a Measure Becomes a Target, It Ceases to Be a Good Measure." Applied to AI: "An AI" "Optimizing" for a "Proxy Goal" "May" "Destroy" the "True Goal" in the "Process." * '''Constitutional AI''' (CAI) β (Anthropic). "A Technique" where an "AI" is "Trained" to "Follow" a **"Set of Principles"** (A Constitution) to "Generate" "Safer Outputs." * '''RLHF''' (Reinforcement Learning from Human Feedback) β (See Article 01). "The Current" "Standard" "Alignment Technique" for "Large Language Models." * '''Interpretability''' β (See Article 607). "The Science" of "Understanding" **"What"** is "Happening" "Inside" an "AI's Neural Networks." * '''The Paperclip Maximizer''' β (Bostrom). "A Famous Thought Experiment": an "AI" "Tasked" to "Make Paperclips" "Converts" **"All Matter in the Universe"** "Into Paperclips" because "Its Goal" has "No" "Stopping Condition." * '''Scalable Oversight''' β "The Problem" of "How" to "Supervise" an "AI" that is **"Smarter Than Any Human Supervisor."** </div> <div style="background-color: #006400; color: #FFFFFF; padding: 20px; border-radius: 8px; margin-bottom: 15px;">
Summary:
Please note that all contributions to BloomWiki may be edited, altered, or removed by other contributors. If you do not want your writing to be edited mercilessly, then do not submit it here.
You are also promising us that you wrote this yourself, or copied it from a public domain or similar free resource (see
BloomWiki:Copyrights
for details).
Do not submit copyrighted work without permission!
Cancel
Editing help
(opens in new window)
Navigation menu
Personal tools
Not logged in
Talk
Contributions
Create account
Log in
Namespaces
Page
Discussion
English
Views
Read
Edit
View history
More
Search
Navigation
Main page
Recent changes
Random page
Help about MediaWiki
Tools
What links here
Related changes
Special pages
Page information