Hardware Podcasts

Can AI revolutionize materials discovery?

Catalyst · 19 Sept 2024 · 37m

What this episode covers

Shayle Khan interviews Ekin Dogus Cubuk, a research scientist at Google DeepMind, about the current state and future potential of using AI for materials discovery. They discuss why historical material discovery was often serendipitous, the limitations of current AI models in predicting novel materials outside known distributions, and the critical need for high-quality experimental data to train these models. The conversation highlights that while AI has not yet yielded commercially deployable materials, it shows promise in optimizing known material classes, such as battery cathodes and superconductors, rather than discovering entirely new paradigms.

Most relevant to

large enterprises

Topics

Mentioned

Companies
Google DeepMind, Microsoft, Energy Hub, Kraken, Anza, Antenna Group, Duracell, Materials Project
Products
Alpha Fold, GNoME, LBCO, YBCO
Also
ICSD

Related episodes from other shows

Other shows on the same ground, closest first.

  • The Manufacturing Executive · 39m

    Why Your Toughest Parts Still Use 50-Year-Old Materials w/ Fazal Mahmood

    Fazal Mahmood, co-founder and CEO of Phase Shift Technologies, explains how AI is accelerating the discovery of new metal alloys to replace 50-year-old materials in industries like oil and gas, aerospace, and defense. He details the critical gap between designing a material and qualifying it for regulated markets, noting that legacy producers often lack the incentive to innovate outside their core franchises. Mahmood highlights three key drivers for this shift: the maturation of additive manufacturing, the performance limits of existing materials in technologies like nuclear fusion, and supply chain disruptions in critical raw materials like tungsten. He illustrates these points with a specific example of a new printable tungsten carbide alloy designed for laser powder bed fusion to solve wear resistance issues in complex flow components.

    Materials & manufacturing processesAI & machine learning
    6 chapters

  • Product Startup · 15m

    209: Materials Research for Consumer Product Design

    Ori Yudilevich, CTO of Materials Zone, explains why materials research is critical for hardware product development, comparing material selection to cooking ingredients where process and quality determine the final outcome. He highlights three dominant trends shaping modern product design: sustainability concerns regarding carbon footprints, supply chain disruptions forcing rapid adaptation to alternative materials, and strengthening global regulations on material safety. Yudilevich describes how his platform, Materials Zone, uses AI and machine learning to model complex material recipes, helping companies reduce time-to-market and costs by predicting outcomes rather than relying on expensive physical testing. The discussion emphasizes the need to balance front-end customer expectations for sustainable, safe products with back-end manufacturing realities like regional supply flexibility and regulatory compliance.

    Materials & manufacturing processesSourcing & supply chain
    7 chapters

  • Oxide and Friends · 1h 10m

    AI, Materials, and Fraud with Ben Shindel

    Ben Shindel, a material scientist and author of the 'BS Detector' newsletter, joins Bryan Cantrill and Adam Leventhal to dissect a retracted economics paper by Aiden Toner Rogers that claimed AI dramatically boosted productivity in materials research. Shindel explains how the paper's data was entirely fabricated, pointing to impossible survey compliance, implausible corporate structures, and suspiciously perfect statistical distributions as evidence of fraud. The hosts compare the incident to Theranos and the work of journalist Steven Glass, discussing how the narrative's appeal to fears about AI's impact on jobs allowed it to bypass rigorous peer review.

    AI & machine learningIntellectual property
    7 chapters

  • The Next Byte Podcast · 23m

    151. Battery Chemistry GPT

    Hosts Daniel and For Bode discuss a University of Michigan project developing a GPT-style AI to design battery electrolytes, aiming to lower EV costs and improve energy density. The team utilizes the Polaris supercomputer and partners with Deep Forest Sciences, adapting drug discovery methods to material science to predict viable electrode-electrolyte pairings. They highlight how this AI-driven approach, combined with automated robotics for testing, could accelerate the discovery of sustainable, affordable battery chemistries for the construction and consumer EV markets.

    Batteries & low-power designAI & machine learning
    7 chapters

  • EMS@C-LEVEL · 17m

    Context, Tools, and Trust:Missing Ingredients in Many AI Strategies, with Luminovo's Timon Ruban

    Timon Ruban, Co-Founder of Luminovo, discusses the evolution of AI in electronics manufacturing, contrasting early hype with current practical applications enabled by large language models. He explains that successful implementation requires balancing comprehensive data context with specialized tools, such as part search APIs, to verify AI proposals. Ruban highlights Luminovo's focus on extracting structured data from unstructured sources like PDFs and emails, and developing 'agentic workflows' where AI assists human decision-making in supply chain management rather than replacing it.

    AI & machine learningSourcing & supply chain
    6 chapters

The summary, chapters and topics are generated from the episode’s audio by this directory, not written by Catalyst. The episode itself belongs to its publisher.