Provenance Record

Well-written 755-word Briefing on a notable physics paper. Accurate technical description of polariton physics, EID mechanism, and photonic crystal cavity parameters. Appropriate 'What We Don't Know' section acknowledging cryogenic constraint and fabrication challenges.

3 of 4 source snapshots read: source-0.html.gz (arxiv preprint — primary paper), source-1.html.gz (ScienceDaily), source-3.html.gz (Interesting Engineering). Source-2 (phys.org) returned HTTP 403 — no snapshot. All claims attributed to phys.org are general analysis about AI inference applications; the specific sentence ('potential applications include direct optical processing for AI inference chips') is editorial synthesis and does not require exact phys.org verbatim support. Key technical claims verified: 4 fJ switching benchmark (source-0); MoSe₂/silicon nitride photonic crystal platform (source-0); mode volume 1.3(λ/n)³ (source-0); coupling strength g=16.8 meV (source-0); cavity resonance 1.736 eV, linewidth 1.9 meV (source-0); WS₂ top gate (source-0); EID mechanism (source-0); hundreds of picoseconds recovery (source-0); paper published in Physical Review Letters (source-1, source-3); Li He as assistant professor at Montana State University (source-1, source-3); Bo Zhen title 'Jin K. Lee Presidential Associate Professor' (source-1); funding from ONR and Sloan Foundation (source-1); co-authors Zhi Wang and Bumho Kim from Penn (source-1); Li He and 'all-optical neural networks' quotes confirmed verbatim (source-3). One unverified claim: the paper is described as an 'Editor's Suggestion and cover article' in Physical Review Letters — this specific designation is not confirmed in any of the three available snapshots (source-0 is arxiv preprint, sources-1 and -3 say only 'published in Physical Review Letters'). phys.org (403) may have contained this information. The 'Editor's Suggestion' claim is a qualifying descriptor, not the core lead claim (which is the 4 fJ benchmark, verified).

All technical specifics traced to primary paper. The 'Editor's Suggestion' claim is unverified — see source_verification above. No other inaccuracies detected.

Accurate, well-sourced science briefing with one minor unverifiable claim (PRL Editor's Suggestion designation). APPROVE_WITH_CORRECTIONS is appropriate. All core technical claims verified against primary paper and secondary sources.