Hardware Podcasts

28. AI Policing Social Media, What Marsquakes Can Teach Us, Preventing Lead Poisoning From Solar Panels

The Next Byte Podcast · 27 Jul 2021 · 22m

What this episode covers

Hosts Daniel and Forbode discuss three engineering topics: Stanford researchers' findings that AI hate-speech detection accuracy drops significantly when trained on human consensus rather than traditional labels; NASA's InSight lander data revealing Mars has a molten core and a single solid crust unlike Earth's tectonic plates; and EPFL Professor Laszlo Forro's solution to prevent lead leakage from perovskite solar cells using transparent phosphate salts.

Most relevant to

large enterprises and established companies

Topics

Mentioned

Companies
Facebook, Stanford University, Apple, Twitter, Instagram, NASA, ETH Zurich, University of Zurich, EPFL
Products
InSight Lander, Perovskite solar cells

Related episodes from other shows

Other shows on the same ground, closest first.

  • Hackaday Podcast · 57m

    Ep078: Happy B-Day MP3, Eavesdrop on a Mars Probe, Shadowcasting 7-Segments, & a Spicy Commodore 64

    Hackaday editors Mike Szczys and Elliot Williams discuss a variety of hardware hacks, including decoding telemetry from the Tianwen-1 Mars probe, a wireless SD card reader using an ESP8266, and a plasma cutter collision detection system based on exact constraint design. They also review a mechanical shadow-casting 7-segment display, a Commodore 64 used to control a meat smoker, and a 4-axis stencil printer for PCBs. The episode concludes with a retrospective on the 25th anniversary of the MP3 format and a discussion on SpaceX's Falcon 9 rapid reflight records.

    Embedded firmwareMechanical engineering
    8 chapters

  • Oxide and Friends · 1h 01m

    AI Detection with Max Spero

    Max Spero, founder and CEO of Pangram, joins Bryan Cantrill and Adam Leventhal to discuss the technology and societal implications of AI detection. They analyze the recent Stanley Druckenmiller Wall Street Journal op-ed as a watershed moment for AI-assisted writing and explain how Pangram's classifier model, Pangram 4, achieves low false positive rates through 'synthetic mirroring' training data. The conversation covers the use of AI detection in hiring processes, the threat of 'humanizer' tools, and the philosophical impact of LLMs on truth and thought.

    AI & machine learningValidation & product-market fit
    6 chapters

  • Hackaday Podcast · 1h 09m

    Ep 249: Data by Laser and Parachute, Bluetooth Hacks, Google's Gotta Google

    Hosts Elliot Williams and Tom Nardi discuss the migration of the Floss Weekly podcast to Hackaday, followed by a deep dive into Gary's reverse engineering of the Google Stadia controller to extract its firmware. They cover NASA's Psyche mission laser communication test sending a cat video, DIY lithium battery upgrades for older Toyota Prius models, and a University of Surrey research paper on generating 3D wire art using diffusion models. The episode concludes with a look at a Raspberry Pi-based custom camera platform and NASA's use of parachute-dropped data cartridges from high-altitude balloons.

    Embedded firmwareAI & machine learning
    10 chapters

  • Hackaday Podcast · 50m

    Ep 350: Damnation for Spreadsheets, Praise for Haiku, and Admiration for the Hacks In Between

    Elliot Williams and Jenny List discuss the loss of NASA's Maven Mars orbiter and review several community hacks, including Cavish Devar's open-source AirPods control software, Abe's lo-fi analog video camera, and Kevin Nokey's Commodore 64 laptop. They also examine a Sudbury Neutrino Observatory update, Pladino's level shifting guide, and Ivan's MIDI slappophone. The episode concludes with Elliot Williams' critique of spreadsheets and Jenny List's praise for the Haiku operating system as a viable daily driver.

    PCB design & layoutIndustrial & consumer product design
    9 chapters

  • Hackaday Podcast · 52m

    Ep009: On the Edge of AI, Comment Your Code, Big Big Wheels, and Makers of Munich

    Hackaday editors Mike Szczys and Elliot Williams discuss new AI-on-the-edge hardware from BeagleBone and Google, the reverse engineering of a keytar firmware, and a critical analysis of Tesla's self-driving safety data. They also cover a custom smart lamp build, flexible PCB contests, and Elliot provides an audio tour of the Make Munich maker fair.

    AI & machine learningDebugging & failure analysis
    8 chapters

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