I also re-organized some tags to make it more even height-wise and still makes sense conceptually, to try and reduce the expansion of the section. We're still three lines shorter than the Rationality section, but that might not last as AI discussion grows.
Additionally, I'd consider merging ChatGPT into the existing GPT tag, but not sure about that.
I've also updated the AI tag. Some of these had already been added there by other helpful people, but I didn't check before adding them directly to the portal.
Christmas wiki editing! Lots of new important tags in the past year.
I added:
Gradient Hacking
Symbol Grounding
AI Success Models
AXRP
Agent Foundations
Eliciting Latent Knowledge (ELK)
Anthropic
Conjecture (org)
AI Persuasion
DALL-E
Multipolar Scenarios
AI Risk Concrete Stories
AI Alignment Fieldbuilding
SERI MATS
AI Safety Public Materials
AI-assisted Alignment
Encultured AI (org)
Shard Theory
Sharp Left Turn
Deceptive Alignment
Apart Research
AI Questions Open Thread
Power Seeking (AI)
AI Alignment Intro Materials
RLHF
Restrain AI Development
I did not add these due to relatively few posts or low strategic importance
Gradient Descent
Spurious Counterfactuals
Tensor Networks
Transformers
Truthful AI
PaLM
Transformer Circuits
EfficientZero
Verification
AI Robustness
2017-2019 AI Alignment Prize
AI Sentience
General Alignment Properties
Adversarial Training
Distributional Shifts
Tiling Agents
Alignment Tax
Auditing Games
AlphaTensor
AlphaStar
PreDCA
Aligned AI Proposals
I'd be open to adding quite a few of them if we're OK with stretching the screen a bit.
This due to it being a super long tag name:
AGI Limits of Engineerable Control & Safety Impossibility Theorems
Currently inactive training programs had lots of tagged posts, but probably won't have many more:
PIBBSS
Refine
These were borderline:
Chain-of-thought Alignment
Compute
I also re-organized some tags to make it more even height-wise and still makes sense conceptually, to try and reduce the expansion of the section. We're still three lines shorter than the Rationality section, but that might not last as AI discussion grows.
Additionally, I'd consider merging ChatGPT into the existing GPT tag, but not sure about that.
I've also updated the AI tag. Some of these had already been added there by other helpful people, but I didn't check before adding them directly to the portal.