Volume 392 | Issue 6805 | 25 June 2026 — Summary of Science publications, focusing on content relevant to AI and life sciences. Note: Issue 6805 is part of Volume 392, not 393 — the numbering doesn't carry a 1:1 relationship with volume across issues.

👁️ COVER STORY — The Molecular Basis of Human Color Vision (Three Companion Papers)

Human daytime color vision depends on three cone opsins — tuned to long (red), medium (green), and short (blue) wavelengths — but until now, structural data explaining their spectral tuning and signaling kinetics has been missing. Three papers in this issue close that gap:

1. Cryo-EM structures of human cone visual pigments — Peng et al. · Science 392, eadz8141 · DOI: 10.1126/science.adz8141
Resolves cryo-EM structures of all three human cone opsins in complex with their G-protein transducin partner, extending the structural picture established for rhodopsin to daylight vision for the first time.

2. Structural insights into spectral tuning and retinal exchange in cone visual pigments — Ohashi et al. · Science 392, eadz3996 · DOI: 10.1126/science.adz3996
Compares red-, green-, and blue-sensitive pigments across catarrhine primates (where red and green opsins arose from a recent gene duplication and differ by only a few residues), providing a structural explanation for how such small sequence differences shift absorption wavelength.

3. Illuminating the molecular basis of human daylight vision (dark-state structures) — Science 392, eadz3624 · DOI: 10.1126/science.adz3624
Determines dark-state structures of the green and blue cone opsins and pairs them with femtosecond-resolution spectroscopy, explaining the fast retinal release/reset kinetics that let cone pigments avoid saturation in bright light — unlike the slower-resetting rod pigment rhodopsin.

Together, these three papers are a strong example of multi-modal structural biology (cryo-EM + ultrafast spectroscopy + comparative primate genomics) resolving a decades-old open question in sensory neuroscience.

🤖 Dartmouth's AI Summer and What Came Next — Kate Crawford

Perspective · pp. 1340–1342 · DOI: 10.1126/science.aei0476

Marking the 70th anniversary of the 1956 Dartmouth workshop that coined the term "artificial intelligence," Crawford (author of Atlas of AI) reflects on the field's wider legacies as today's AI boom collides with material reality: hyperscale data centers creating heat islands near the Great Salt Lake, hundreds of billions of dollars in projected AI infrastructure spend, labor-market disruption forecasts, and expanding military AI contracts. A sobering, non-technical companion piece for anyone tracking the societal footprint of the current AI buildout.

Science

🍽️ Time-Restricted Feeding Extends Healthspan (and, in Males, Lifespan) — In Other Journals summary

In Other Journals · pp. 1353–1354 · DOI: 10.1126/science.aej9373

Summarizing Iiams et al., this piece covers a mouse study on healthy (not just metabolically diseased) animals: restricting feeding to 12- or 8-hour daily windows improved health markers in both sexes, with the 8-hour window (which also drove voluntary caloric restriction) showing the strongest effects — though lifespan extension was only observed in male mice. Relevant context for the growing "AI + longevity biology" data landscape Bioinfo.AI has been tracking.

Science

⚡ Superconducting Phase Diagram of Multilayer Square-Planar Nickelates — Pan et al.

Research Article · pp. 1390–1395 · DOI: 10.1126/science.adp4440

By varying the number of NdNiO₂ layers sandwiched between NdO₂ spacer layers, the team synthesized a series of nickelate compounds spanning a wide range of superconducting behaviors — including cuprate-like physics at low layer counts — offering a new experimental platform for probing high-temperature superconductivity mechanisms. Materials-science content rather than life sciences, but relevant if Bioinfo.AI ever covers AI-accelerated materials discovery.

Science

🏛️ Policy Watch: NSF Slashes Research Programs to Fund New Tech Initiative

News Feature · By Jeffrey Mervis

Reporting that NSF is cutting budgets for hundreds of traditional basic-science programs by 20–30%, despite an overall budget cut of just 3%, apparently to redirect funds toward a new $1.5 billion initiative aimed at commercializing NSF-funded research. Relevant backdrop for anyone tracking U.S. federal science funding trends alongside the NIH grant-pacing story covered in the following week's issue (6806).

Science

Issue theme in brief: This is a structural-biology-heavy issue, headlined by the color-vision opsin trio — a rare case of three independently authored, mutually reinforcing papers resolving one open question together. The clearest AI-angle piece is Kate Crawford's Dartmouth anniversary Perspective, which is more a societal/policy reflection than a technical AI paper, so if Bioinfo.AI wants a technical AI/bioinformatics anchor for this week, the opsin structures (particularly the deep structural basis for spectral tuning) are the stronger fit.