explore | Grilled Cheese

Explore Log in Sign up

Terms of Use Privacy Policy Community Standards Help Get the app

Grilled Cheese is a product of Village Compute

Version devBuilt at: 2026-10-17 11:32:24 EDT

Photos Videos All People

@softwarecampus.bsky.social Oct 8, 2026, 1:43 PM

DeeBOP: AI for biological sequence design. 🧬 Software Campus participant Erik Kubaczka at the #TUDarmstadt works with #Merck on Bayesian optimisation for high-throughput biology. 🔬🤖 Read: softwarecampus.de/en/projekt/d...

#SoftwareCampus #AIforScience #BioAI

Photo, no description

@artificialscience.bsky.social Oct 8, 2026, 5:43 AM

AI 'lab teams' now propose, debate and test hypotheses. Some of it has held up at the bench; some has not. Here is how multi-agent AI labs really work. Read the breakdown. #AIforScience #MultiAgentAI #Biotech #AgenticAI

artificialscience.org/2026/10/mult...

@antifocal.bsky.social Oct 7, 2026, 4:22 PM

When neighbouring elements scatter X-rays alike, ordered and disordered phases give near-identical powder patterns. People chose the targets and in the end re-analysed the patterns by hand.

Szymanski et al., Nature 624, 86 (2023)

Values are representative.

#MLSky #AIforScience #research

@antifocal.bsky.social Oct 7, 2026, 4:20 PM

Patchy films gave noisy readings, so each sample ran two to four times. People set the ranges, picked the scale-up recipe and saw a DMSO test film form defects at 100 cm².

Wang et al., Nature Communications 16, 1498 (2025)

Values are representative.

#MLSky #AIforScience #research

@aidailypost.com Oct 6, 2026, 5:41 PM

OpenAI just set up an advisory group to tackle the power surge behind its AI models—think Navier‑Stokes, math breakthroughs, and open‑source GitHub projects. Curious how this could reshape the scientific community? Dive in. #OpenAIAdvisory #AIForScience #NavierStokesAI

🔗

Photo, no description

@cairninstitute.bsky.social Oct 5, 2026, 4:42 PM

Plants invented entire regulatory families animals never evolved. That's visible in 2 seconds on this heatmap.
Free for academic use. From CAIRN Institute.
#Genomics #TranscriptionFactors #PlantScience #ComparativeGenomics #AIforScience

Photo, no description

@antifocal.bsky.social Oct 5, 2026, 12:41 PM

When neighbouring elements scatter X-rays alike, ordered and disordered phases give near-identical powder patterns. People chose the targets and in the end re-analysed the patterns by hand.

Szymanski et al., Nature 624, 86 (2023)

Values are representative.

#MLSky #AIforScience #research

@antifocal.bsky.social Oct 4, 2026, 7:37 PM

People set the safe voltage range, chose the two plunger gates, designed the score and labelled every map. This was coarse tuning, not a qubit, with five runs per device.

Moon et al., Nature Communications 11, 4161 (2020)

Values are representative.

#MLSky #AIforScience #research

@sciencecareers.bsky.social Oct 3, 2026, 11:21 PM

The Chen Institute & @science.org Prize for AI Accelerated Research is open for entries — recognizing researchers using AI to make breakthroughs in physical & life sciences.

🏆 Up to $30K + Get published in Science.

Apply now: https://bit.ly/4gWBaxu

#AIforScience #MachineLearning

Promotional image for the Chen Institute & Science Prize for AI Accelerated Research. It invites early-career scientists to submit essays using AI to advance science, with a "Start application" button. The background gradient transitions from blue to orange.

@antifocal.bsky.social Oct 3, 2026, 1:59 PM

People still polish the surface, set voltage, tilt and detector distance, then check any phase the model has not seen.

Kaufmann et al., Science 367, 564 (2020)

Values are representative.

#MLSky #AIforScience #research

@antifocal.bsky.social Oct 3, 2026, 1:55 PM

The detector loses the phase, so different profiles can fit one curve. The operator set bounds for every parameter. In a PTCDI-C₈ test, results below about 250 Å scattered more.

Pithan et al., J. Synchrotron Rad. 30, 1064 (2023)

Values are representative.

#MLSky #AIforScience #research

@tenuretracker.bsky.social Oct 2, 2026, 7:09 AM

Harvard University (@harvard.edu) is hiring:
Tenured Professor in Applied Mathematics and Statistics

#fullprofessor #aiforscience #appliedmathematics

@sciencecareers.bsky.social Sep 30, 2026, 10:51 PM

The Chen Institute & @science.org Prize for AI Accelerated Research is open for entries — recognizing researchers using AI to make breakthroughs in physical & life sciences.

🏆 Up to $30K + Get published in Science.

Apply now: https://bit.ly/4gWBaxu

#AIforScience #MachineLearning

A person in a lab coat and gloves holds a petri dish up for examination. Text reads, "Get published in Science. Now accepting essays from early‑career scientists using AI to advance science." Logos for Chen Institute & Science Prize for AI Accelerated Research are displayed. A "Start application" button is visible.

@cairninstitute.bsky.social Sep 29, 2026, 10:43 PM

The goal is not a fluent black-box answer. It is an inspectable route from a research question to an atlas-grounded hypothesis, evidence boundary, and next experiment.

www.cairninstitute.com/blogs/Darwin...

#Genomics #GeneRegulation #Bioinformatics #AIforScience #PlantScience

@antifocal.bsky.social Sep 29, 2026, 4:27 AM

🔬 𝗖𝗮𝗻 𝗮𝗻 𝗔𝗜 𝗮𝗹𝗶𝗴𝗻 𝗮𝗻 𝗲𝗹𝗲𝗰𝘁𝗿𝗼𝗻 𝗺𝗶𝗰𝗿𝗼𝘀𝗰𝗼𝗽𝗲 𝗯𝘆 𝗹𝗼𝗼𝗸𝗶𝗻𝗴 𝗮𝘁 𝘁𝗵𝗲 𝗶𝗺𝗮𝗴𝗲 𝗮𝗹𝗼𝗻𝗲?

Bayesian optimisation tuned a STEM on image contrast: 18 to 32 images per aberration.
Amorphous or drifting samples defeat it. People set region and bounds.

Values are representative.

#MLSky #AIforScience #research

Infographic titled AI tuned a STEM's lenses, cutting astigmatism from 12 to 2 nm. Subtitle: Bayesian optimisation changed the corrector settings and scored the contrast of each new image. A loop of four stations, 30 to 40 s per turn. 1, Set: lens and dial, change corrector settings. 2, Image: atom columns, scan the sample. 3, Score: a wide histogram marked sharp beside a narrow one marked blurred, contrast as variance. 4, Choose: a model curve with a dashed line at its peak, pick next setting. Centre: 18 to 32 images per aberration. People band: chose a clean, high-contrast region, set the search bounds per sample, picked the noise model, left third order to standard software. Figures: astigmatism 12 to 2 nm, checked by ptychography; axial coma 554 to 224 nm; 18 to 25 min per run against 68 s for a standard tableau. Bold line: The loop only works where the sample gives the image contrast worth sharpening. Source: Pattison et al., npj Comput. Mater. 11, 274 (2025).

@kejunying.bsky.social Sep 28, 2026, 5:13 PM

10/10 Thank you to all our collaborators, including @jameszou.bsky.social and the entire HealthFlux team!

#AIforScience #AIforAging #WorldModels

@artificialscience.bsky.social Sep 28, 2026, 5:17 AM

AlphaFold just released free 3D structures for 2,800+ human-infecting viruses to support pandemic preparedness. Any lab on Earth can download them.

#AlphaFold #PandemicPreparedness #AIforScience #Virology #DeepMind

artificialscience.org/2026/09/alph...

@science-pubmed.bsky.social Sep 28, 2026, 1:54 AM

🤖 Free master prompt: paste into ChatGPT / Claude / Gemini → your AI becomes a PubMed paper-chat assistant. Grounded strictly in the abstract you paste. Grad students, postdocs, PIs welcome. Unlimited use in your own subscription. https://sciencepubmed.net/en/lab/toolkit/ #AIforScience #PhDchat

@antifocal.bsky.social Sep 26, 2026, 2:28 PM

🔬 𝗗𝗼𝗲𝘀 𝗕𝗮𝘆𝗲𝘀𝗶𝗮𝗻 𝗼𝗽𝘁𝗶𝗺𝗶𝘀𝗮𝘁𝗶𝗼𝗻 𝗸𝗻𝗼𝘄 𝘄𝗵𝗲𝗿𝗲 𝘁𝗵𝗲 𝗯𝗲𝘀𝘁 𝗿𝗲𝗰𝗶𝗽𝗲 𝗶𝘀?

Tested points plus uncertainty make a map; the next run goes where the score peaks.
Engineer first, algorithm last: $52,000 to target vs $105,000 alone.

Figures as reported.

#MLSky #AIforScience #research

Diagram titled How the next experiment gets chosen. The upper plot, The map, shows five tested points, a predicted curve and an uncertainty band, wide in the untested gap. The lower plot, The score, peaks inside that gap at the point marked next experiment. Labels: Exploit, go where the prediction is high. Explore, go where the band is wide. Card one, Benchmark of 5 materials datasets: finding 80% of the top 5% of 600 printed structures takes about 75 samples with one lengthscale per variable, 85 with a random forest model and 170 with one lengthscale for all. Card two, Plasma etch game with 11 parameters, simulated process: 105,000 dollars to target for a senior engineer alone, a median 52,000 dollars for engineer first, algorithm last. Dark bar: up to 8 times faster than random sampling; 13 of 300 algorithm-only runs beat the engineer; half the cost. Footer: People are strongest early in the search; the algorithm is strongest near a tight target.

@antifocal.bsky.social Sep 24, 2026, 5:16 AM

🔬 𝗜𝘀 𝗮 𝗱𝗮𝘁𝗮 𝗳𝗶𝗹𝗲 𝗲𝗻𝗼𝘂𝗴𝗵 𝗳𝗼𝗿 𝗮𝗻 𝗔𝗜 𝘁𝗼 𝗿𝗲𝘂𝘀𝗲 𝘆𝗼𝘂𝗿 𝗺𝗲𝗮𝘀𝘂𝗿𝗲𝗺𝗲𝗻𝘁?

Without the X-ray wavelength, a 30° peak could be 2.98 Å or 1.37 Å.
The odds that a paper's data still existed fell 17% per year.

Figures as reported.

#MLSky #AIforScience #research

Chart titled The file keeps the numbers. The context stays behind. Three panels sit side by side. Wavelength: one XRD peak at 2θ = 30° labelled Cu Kα? and Mo Kα?, with the wavelength not in the file. Callout: 2.98 Å or 1.37 Å, since Cu Kα at 1.5406 Å gives 2.98 Å while Mo Kα at 0.7107 Å gives 1.37 Å. Scale: a sketch of round particles with one diameter marked 56 px = ? nm, with the pixel size not in the file. Callout: 10 to 28% of 580 life-science papers gave no scale information for their images. Survival: a sketch of two falling curves for data and email against article age, for 516 papers 2 to 22 years old. Callout: the odds the data still existed fell 17% per year; a working author email fell 7% per year. A dark bar lists what should travel with the file: bibliographic, specimen, instrument and image data. Footer: a model can only reuse what the file remembers.