AI-Assisted Materials Research with the Atlas

1. Learning objectives

• Explain why source-grounded AI is needed.
• Use claim-evidence matrices as RAG-ready material.
• Draft a literature-to-mixture reasoning workflow.

2. Key concepts

• Wiki-RAG
• Source-grounded AI
• Claim-evidence matrix
• Knowledge graph
• Mixture recommendation
• Experiment feedback loop

3. Minimal theory

Generative AI should retrieve and combine grounded evidence, not invent mixture rules. The Atlas provides book anchors, paper evidence, graph edges, and lab positioning.

4. Source-grounded evidence

• AI positioning: 06_lab_position/materials_ai_wikirag_position.md.
• Workflows: 09_ai_workflows/wiki_rag_ingestion_plan.md, 09_ai_workflows/atlas_query_templates.md.
• Graph files in 07_visualization/.

5. Representative papers

• Atlas claim-evidence matrices.
• Green ECC synthesis node.
• Extreme ductility synthesis node.
• Lab positioning cards.

6. Figures/tables to show later

• RAG architecture diagram.
• Claim-evidence row structure.
• Query routing example.

7. Discussion questions

1. What can AI safely infer from source notes?
2. What requires PDF-level verification?
3. How do failed experiments return to the Atlas?

8. Assignment idea

Write three RAG queries and identify which Atlas files should be retrieved.

9. Linked Atlas nodes

• 06_lab_position/materials_ai_wikirag_position.md
• 09_ai_workflows/wiki_rag_ingestion_plan.md
• 09_ai_workflows/atlas_query_templates.md
• 07_visualization/graph_paper_to_node_edges.csv

10. Suggested reading path

1. 06_lab_position/materials_ai_wikirag_position.md
2. 07_visualization/claim_evidence_classification_summary.md

Evidence files

• 07_visualization/foundational_papers_claim_evidence_matrix.csv
• 07_visualization/extreme_ductility_extension_claim_evidence_matrix.csv
• 07_visualization/sustainable_ecc_extension_claim_evidence_matrix.csv
• 07_visualization/by_lee_lab_publications_claim_evidence_matrix.csv