AI Safety Explained

Home
Archive
About
New
When Might we Expect Artificial General Intelligence? Why?
An AI that is able to do everything a human is able to do seems far off. But when do AI experts think this will be developed, and why?
 • 
Victoria Brook
A Safer Form of AI?
'Quantilizers' are a proposed form of AI that selects randomly from the most effective e.g. 10% of actions, rather than taking the most effective…
 • 
Victoria Brook
What is OpenAI's approach to making AI Safe?
OpenAI, developer of ChatGPT and DALLE-2, are a big name in AI. But what are they doing to make sure that AI are safe, now and in the future?
 • 
Victoria Brook
How 'safe' AI can still have problems
Mesa-optimization is an important concept in AI alignment. It can make sufficiently intelligent AI difficult or impossible to control or make safe.
 • 
Victoria Brook
AI Safety Explained
AI Safety Explained
AI has the potential to do serious harm over the next 5-10 years, but understanding AI thoroughly is difficult. This page makes it easy by explaining basic concepts and papers in AI safety in simple terms.
Recommendations
Astral Codex Ten
Astral Codex Ten
Scott Alexander
Million Year View
Million Year View
Riley Harris
AI Safety Newsletter
AI Safety Newsletter
Center for AI Safety

AI Safety Explained

AboutArchiveRecommendations
© 2023 Victoria Brook
Privacy ∙ Terms ∙ Collection notice
Start WritingGet the app
Substack is the home for great writing