home.social

#aiinscience — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #aiinscience, aggregated by home.social.

fetched live
  1. 🧬 Can we trust membrane protein annotations when databases tell different stories?

    🔗 MetaMP Ecosystem for Unified, Auditable, and Benchmark-Ready Data for Reliable Membrane Protein Annotation. Computational and Structural Biotechnology Journal (CSBJ). DOI: doi.org/10.34133/csbj.0165

    📚 CSBJ - A Science Partner Journal: spj.science.org/journal/csbj

    #MembraneProteins #StructuralBiology #Bioinformatics #ComputationalBiology #ProteinScience #DataScience #DrugDiscovery #AIinScience #OpenScience

  2. Terrence Tao, armed with his infinitely superior intellect, tries to tackle the Jacobian Conjecture with... #ChatGPT. 🤖💡 Because obviously, the best way to solve a decades-old math problem is to have a digital assistant rephrase its terms of service while asking you to log in. 🙄🔐
    chatgpt.com/share/6a5fdc7a-d6f #TerrenceTao #JacobianConjecture #MathHumor #AIinScience #HackerNews #ngated

  3. Generative AI may be changing not only how we write scientific papers, but also who we write them with.

    A fascinating new #preprint by Akira Matsui finds that the decades-long decline in solo authorship halted, and in many fields reversed, after the release of ChatGPT.

    :oa: arxiv.org/abs/2607.10780

    A provocative hypothesis: AI may be taking over some of the work previously done by human co-authors.

    #GenAI #Authorship #ScholarlyCommunication #AIinScience #ScienceOfScience

  4. MatterChat: Bridging the Gap Between Atomic Structures and Large Language Models

    Scientists made MatterChat, an AI that connects language models with atomic physics. It helps predict material properties.

    #MatterChat, #AIinScience, #AtomicPhysics, #LLM, #LawrenceBerkeleyLab

    newsletter.tf/matterchat-ai-un

  5. New Chemical Discoveries Pursued Through Lab and Digital Tools

    Caltech scientist Melissa Ramirez uses computers and robots to create new spirocycle molecules for potential medicines. Learn how technology speeds up drug discovery.

    #DrugDiscovery, #Chemistry, #Caltech, #AIinScience, #Robotics

    newsletter.tf/caltech-scientis

  6. Virginia Tech published something many universities still lack: Practical guidance on how to use AI responsibly in research - step by step, across the entire research lifecycle.

    🔗 research.vt.edu/research-suppo

    Not abstract principles. Concrete prompts. Clear warnings about IP, confidentiality, hallucinations, and dual-use risks. #AI should support thinking — not replace it.

    #ResponsibleAI #ResearchIntegrity #OpenScience #AIinResearch #AIinScience #AIethics #universities

  7. A slightly tongue-in-cheek AI animation of a conceptual model showing leakage of highly mineralized mine waters into a carbonate aquifer:
    youtu.be/WDIze8QNMYs

    Yes - it’s simplified. Yes - real systems are messier.
    But sometimes a "clean" visual helps to communicate density-driven flow and reactive processes more intuitively than 20 pages of equations.

    The animation idea emerged while working on my monograph. Draft versions are available here:
    zenodo.org/records/16741148

    Engineering humor is allowed - even in hydrogeochemistry 😁

    #Hydrogeochemistry #Groundwater #MineWater #ReactiveTransport #CarbonateAquifer #PHREEQC #AIinScience #Geoscience #SvystunovaGully

  8. For ten years, #CRediT 🎂 has offered 14 standardized contributor roles that make authorship transparent: nature.com/articles/d41586-025

    But there is a clear gap: CRediT has no place for #AI tools. So how should we properly disclose the use of ChatGPT & Co in research? One answer is #GAIDeT - Generative AI Delegation Taxonomy w3id.org/gaidet

    #ResearchIntegrity #Responsible #OpenScience #Infrastructure #AIinResearch #Metadata #ResearchAssessment #AIinScience #AIethics

  9. AI is reshaping molecular visualization: real‑time 3D, natural‑language interaction, predictive modeling, and interactive simulations. This could democratize workflows from protein folding to drug discovery—months to hours. Anyone building or using unified platforms for macromolecules? What’s working, what’s missing? 🤖🧪
    #MolecularVisualization #AIinScience #DrugDiscovery #OpenScience
    is.gd/Ss5KUO

  10. Make #AI use transparent & machine-readable. 🙌 We’re looking for journal editors ready to pilot GAIDeT in their submissions - even a few DOIs can set a new standard!

    It’s simple to try - and if you don’t believe it, check the instruction: 👉 doi.org/10.5281/zenodo.17101228

    #Crossref #AIinScience #OpenScience #ScholComm #AIethics

  11. AI has already become a routine part of research practice, but a simple disclosure like “we used ChatGPT” does not explain what exactly the #AI did. A new post in Research Professional News presents another approach – the Generative AI Delegation Taxonomy (GAIDeT), which clearly records which tasks, at which stages of research, were delegated to AI, ensuring transparency and trust.

    👉 researchprofessionalnews.com/r

    #OpenScience #ResearchIntegrity #AIinScience #Transparency

  12. In a study published in the journal Cell, researchers detailed how they analyzed 36 million chemical compounds using neural networks. Read the study: cell.com/cell/abstract/S0092-8
    #CellJournal #Research #NeuralNetworks #ChemicalCompounds #AIinScience

  13. JOB: W2-Forschungsprofessur für Angewandtes Maschinelles Lernen für Naturwissenschaft und Technik an der Hochschule München.
    Start: Sommersemester 2026 oder später
    Schwerpunkte: u. a. Deep Learning, Physik-Informatik, Robotik, Photonik, Bioinformatik, Materialinformatik.
    Mit forschungsstarkem Fokus und reduzierter Lehrverpflichtung (9 SWS).
    Bewerbung bis 12.09.2025
    #MaschinellesLernen #Professur #AIinScience

    academics.de/jobs/forschungspr

  14. 🤖 You don’t need to code in Python to run machine learning analyses.
    Galaxy includes a full suite of ML tools, from classification to clustering to cross-validation, all in your browser.
    Use real datasets. Build real models. Share your work.
    Explore machine learning tools on usegalaxy.org

    #MachineLearning #DataScience #GalaxyProject #AIinScience