AIMATRY AI-native scientific-institution source packet v0.1

[hummbl-dev]

Parent and dependencies

• Standing research ledger: https://github.com/hummbl-dev/hummbl-research/issues/134
• Source-authority registry: AIMATRY / AI-native materials institutions bibliography and source-authority registry v0.1 hummbl-bibliography#123
• Claim-first science–academia separation: https://github.com/hummbl-dev/hummbl-research/issues/106

Status

CANDIDATE PUBLIC-SOURCE RESEARCH PACKET — CLOSEABLE — NON-CANONICAL

Purpose

Create the bounded, reviewable evidence packet supporting the AIMATRY and AI-native scientific-institution case study.

The packet must preserve exact source identity, dates, claim scope, non-support boundaries, asset-specific evidence, institutional self-description, independent corroboration, contradictions, corrections, and falsifiers. It is the evidence substrate for the standing ledger and institutional comparison; it is not itself the final geopolitical or architectural conclusion.

Required claim families

Keep these separate:

1. Appointment and affiliation — what position changed, when, and with what retained affiliations.
2. Institutional mandate — what AIMATRY is officially intended to build.
3. Technical lineage — which AI/materials capabilities were demonstrated before the move.
4. Asset transfer — which people, methods, data, models, software, laboratories, IP, companies, contracts, production systems, standards, or curricula actually moved, replicated, stayed, or remain unknown.
5. Performance claims — stated cycle-time targets versus measured outcomes.
6. Institutional comparison — AIMATRY, Berkeley BIDMaP, and DOE Genesis Mission on common dimensions.
7. Funding and policy context — documented conditions without asserting unsupported singular causality.
8. Standards power — evidence of lifecycle standards activity or adoption.
9. Commercialization and production — discovery-to-scale evidence.
10. Negative, corrective, and null evidence — evidence weakening the capability-transfer thesis or correcting prior reports.

Seed source set

Primary and authoritative institutional sources

• Tsinghua appointment and AIMATRY description
  https://www.tsinghua.edu.cn/en/info/1244/14984.htm
• UC Berkeley current faculty record
  https://chemistry.berkeley.edu/people/omar-yaghi
• Nobel Prize in Chemistry 2025
  https://www.nobelprize.org/prizes/chemistry/2025/press-release/
• Berkeley BIDMaP institutional description
  https://bidmap.berkeley.edu/
• U.S. Department of Energy AI / Genesis Mission
  https://www.energy.gov/topics/artificial-intelligence

Reporting and correction

• Nature report plus July 9, 2026 clarification
  https://www.nature.com/articles/d41586-026-02143-x

Technical lineage

• ChatGPT Chemistry Assistant for Text Mining and Prediction of MOF Synthesis
  https://arxiv.org/abs/2306.11296
• A GPT-4 Reticular Chemist for Guiding MOF Discovery
  https://arxiv.org/abs/2306.14915
• Image and Data Mining in Reticular Chemistry Using GPT-4V
  https://arxiv.org/abs/2312.05468
• System of Agentic AI for the Discovery of Metal-Organic Frameworks / MOFGen
  https://arxiv.org/abs/2504.14110
• Algorithmic iterative reticular synthesis of zeolitic imidazolate framework crystals / AIRES
  https://www.nature.com/articles/s44160-025-00939-9
• Large language models for reticular chemistry
  https://www.nature.com/articles/s41578-025-00772-8

Translation, safety, and commercialization constraints

• Make metal–organic frameworks safe and sustainable by design for industrial translation
  https://www.nature.com/articles/s41578-025-00774-6
• A market for metal–organic frameworks
  https://www.nature.com/articles/s41563-025-02147-4
• Metal–organic frameworks for the future
  https://www.nature.com/articles/s41565-025-02095-9

Evidence-unit requirements

Reuse the repository's current packet schema where possible. Each evidence unit should expose at least:

packet_claim_id:
bibliography_key:
source_locator:
source_revision_or_version:
publication_date:
event_or_effective_date:
retrieved_at:
source_authority_role:
primary_or_secondary:
institutional_self_report: true | false
claim_family:
asset_class:
comparison_target: AIMATRY | BIDMaP | DOE_GENESIS | YAGHI_TECHNICAL_LINEAGE | OTHER
exact_capability_or_event_demonstrated:
relevant_location:
claim_posture:
source_can_support:
source_cannot_support:
transfer_posture: retained_us | moved | replicated | jointly_accessible | newly_created_china | newly_created_us | unclear | no_transfer_evidence
performance_posture: target | author_reported | independently_measured | not_applicable | unknown
publication_maturity:
contradiction_or_correction_link:
falsifier_or_revision_trigger:
rights_or_quote_boundary:
notes:

Avoid long copyrighted excerpts. Prefer precise locations, concise compliant excerpts, and source-faithful paraphrases.

Required packet lanes

Lane A — Appointment and affiliation timeline

Record at minimum:

• Tsinghua's July 3, 2026 appointment event;
• full-time language;
• Berkeley's current emeritus / Professor of the Graduate School title;
• retained public Berkeley contact, lab, institute, or company references without treating stale biography text as current affiliation proof;
• prior Tsinghua honorary-professor relationship;
• unresolved dual-affiliation or transition details.

Lane B — AIMATRY institutional mandate

Extract and classify the official claims concerning:

• disciplines and departments;
• materials design and synthesis;
• full chain from theory/computation to R&D/production;
• order-of-magnitude cycle-compression target;
• lifecycle technical standards;
• teaching and talent formation.

Mark each as a stated goal unless measured evidence exists.

Lane C — Technical lineage

Build evidence units showing the progression from:

literature and image extraction
→ structured experimental memory
→ prediction
→ human–AI experimental iteration
→ agentic generation and simulation
→ synthesis-feasibility filtering
→ automated synthesis and characterization
→ algorithmic closed-loop experiment selection
→ translation and production constraints

Preserve peer-review and preprint maturity for each step.

Lane D — Asset-separation ledger

Populate the asset classes defined in hummbl-research#134. A blank or unknown record is preferable to an inferred transfer.

Lane E — Institutional comparison

Produce evidence units supporting the common AIMATRY ↔ BIDMaP ↔ DOE Genesis comparison dimensions in hummbl-research#134.

Lane F — Capability-transfer tier

Support or reject tiers T0–T8 from the parent issue. Record the highest evidence-supported tier and the missing evidence required for the next tier.

Lane G — Funding and causality

Collect primary or high-authority evidence on relevant U.S. research-funding and institutional-policy conditions. Separate:

• funding request;
• enacted funding;
• program-specific changes;
• grant cancellations or delays;
• researcher statements;
• inferred motivation;
• direct causal statements.

Do not infer that contextual timing proves causation.

Lane H — Standards, production, and commercialization

Track concrete standards documents, committees, specifications, patents, licenses, industrial partnerships, pilot production, manufacturing scale, field deployments, and independently observed performance.

Adversarial checks

The packet must reject or flag:

• a full-time appointment treated as proof of citizenship change;
• institution self-description treated as independent execution evidence;
• a 10× target represented as achieved performance;
• a person's relocation treated as automatic transfer of IP, companies, datasets, models, laboratories, or contracts;
• a preprint presented as peer reviewed;
• author-reported synthesis or performance presented as independent replication;
• Berkeley legacy biography text treated as current active-role proof without reconciling the title field;
• funding-policy context presented as the sole cause without a direct source;
• prestige, citation count, or Nobel status treated as proof of implementation success;
• absence of public evidence represented as proof that an asset did or did not transfer;
• press coverage presented as a primary source for institutional details already available from official records.

Required outputs

1. Packet manifest with stable/versioned sources.
2. Claim map for every material statement in hummbl-research#134.
3. Appointment and affiliation timeline.
4. Technical-lineage evidence map.
5. Asset-separation ledger seed.
6. AIMATRY ↔ BIDMaP ↔ DOE Genesis evidence matrix.
7. Capability-transfer tier assessment with evidence gaps.
8. Correction, contradiction, null, and uncertainty register.
9. Rights and quotation manifest.
10. Valid and adversarial/invalid packet fixtures where repository conventions support them.
11. Bounded implementation receipt.

Acceptance criteria

• Every current material claim in hummbl-research#134 resolves to packet evidence or is marked unresolved.
• Bibliography authority references hummbl-bibliography#123 rather than copying canonical source metadata.
• Publication, event, effective, revision, and retrieval dates remain distinct.
• Tsinghua, Berkeley, Nobel, DOE, and BIDMaP self-descriptions are clearly marked as institutional sources.
• Nature's clarification is accurately represented.
• Peer-reviewed AIRES and preprint MOFGen remain distinct.
• AIMATRY's stated 10× target is separated from measured results.
• Personnel, trainee network, methods, data, models, software, labs, compute, IP, companies, contracts, production, standards, curriculum, and funding are tracked separately.
• Negative, corrective, contradictory, null, and missing evidence is retained.
• The packet permits independent review without relying on chat history.
• No geopolitical, legal, espionage, patent, export-control, or national-security conclusion is made beyond the evidence.
• No GitHub-hosted Actions minutes are consumed.

Non-goals

• Writing the final narrative analysis.
• Declaring a national winner.
• Contacting institutions, authors, companies, funders, or standards bodies.
• Downloading or executing external software or model weights.
• Creating deployments, experiments, purchases, or outreach.
• Promoting a HUMMBL/BaseN/Ownward institutional model to canon.

Close condition

Close when the first complete packet is committed, validates against the repository's source-packet contract, covers all material claims in hummbl-research#134, and is independently reviewable.

[hummbl-dev]

---

Automated triage run (2026-07-24)

• Category: research
• Priority: P3
• Severity: low
• Staleness: fresh
• Recommended next action: Assemble the bounded evidence packet under packets/ and link to the parent research ledger hummbl-research#134.

Notes: Candidate packet for AI-native scientific-institution capability transfer; needs source locators and claim map before closeable.

[hummbl-dev]

Cross-reference (2026-08-06): This issue overlaps with https://github.com/hummbl-dev-org/hummbl-research/issues/134.

Reason: Both track AIMATRY AI-native scientific institution capability transfer.

Recommended action: Determine which repo is the canonical home for this work, link the two issues, and consider closing the duplicate. This is a read-only triage comment — no status change made.