Next Release of the AI Alignment Research Graph

[mnm-matin]

Just one issue to track progress till the next release and collaborate on creating the right issues. Feel free to edit this issue/comment on changes.

Scope of Next Release:

• For each paper create a .md and link it to other relevant papers (i.e connected papers)
• Provided a summary for each paper
• An intro for people new to AI Safety, simmilar to https://aisafety.info/
• Chat with paper (i.e similar to elicit)?

LLM Clustering:

• LLM Related Tasks #12
• solve Hierarchical Clustering  #13
• problem: current approach only scalable to a fixed number of papers equal to context window, likely need the LLM to provide a diff instead of re-writing the whole file
• also relevant as seed/evaluation AI safety Landscape

Dataset:

• Need to write a crawler/scrapper to get data directly from arxiv (@mnm-matin will push a small prototype script)
• Need to be able to search/filter for AI Safety papers on arxiv
• Download papers/extract abstract from those papers and any relevant tags
• Cron job to download new published paper or github-actions
• extract citations for each paper?
• extend to non-arxiv papers, this project

Frontend

• Frontend improvements #15
• Make it look more like connectedpapers.com/ Visual Prompting Elements
• Make search more prominent
• Reduced useless white space on the right
• Add "New to AI Safety Button"

Misc Tasks

• Name: Should we rename (is it AI Alignment Graph? AI Alignment Research Graph? Ai Safety? PapersGraph?)
• Move the repo to a github organization (organization name = AI meets Safety?)
• Squash commits in the repo while maintaining upstream sync
• Discord Community
• Setup weekly recurring public meeting
• Contributors Graph in the README?

Funding Opportunities

• EA Grants, Open Philanthropy
• Microsoft Founders Hub (£150k in OpenAI credits)

AI Research Tools:

• OpenKnwoledgeGraph
• ConntectedPapers - citation based graph
• visualizing scientific landscapes (vosviewer) - java based
• ResearchRabbit.ai
• lateral.io
• scite.ai
• https://aisafety.info/ - good intro to AI Safety
• https://influencemap.cmlab.dev/ - visualizes citation influences among academic entities
• semanticscholar.org - graph construction in semantic scholar

Potential Collaborators:

• https://ai-plans.com/

[mnm-matin]

Need two names one for repo and one for github org:
For github org:

• ai-meets-safety
• ai-safety-network
• ai-alignment-hub

For Repo:

• AlignmentGraph
• AIAlignmentGraph
• SafetyGraph
• AISafetyNavigator

here are some LLM generated pairings:

• ai-safety-explorers/SafetyMapAI
• ai-alignment-community/AlignmentAtlas
• ai-safety-network/SafetyWebViz
• ai-research-compass/AIRiskNavigator
• ai-meets-safety/SafetyLensAI
• ai-alignment-hub/AlignmentArchive
• ai-safety-connectors/SafetyNexusGraph