Oxide and Friends

Nicholas Carlini discusses adversarial machine learning, revealing how sequences of tokens can trick language models into ignoring restrictions. The hosts explore the peculiarities of C programming and delve into the surprising effectiveness of adversarial attacks on machine learning models, emphasizing the need for security-conscious approaches in ML development.

Charlie Park, Oxide colleague, shares his experience with data visualization. Topics include flame graphs, Minard / Tufte diagrams, DTrace aggregations, ParaView, Simpsons IMDb analysis, survivorship bias, and Kartlytics. The hosts also discuss gnuplot, OODA loop, and upcoming Oxide and Friends book club.

Members of the Oxide storage team, including Josh, Alan, James, and Matt, discuss Crucible, a block storage service. They talk about the challenges of storing data, building a crash-safe disk representation, starting a new project, and collaboration. They also cover testing system reliability, resolving back pressure issues, and the development and open-sourcing of Crucible.

Steve Klabnik joins the podcast to discuss various topics including innovation stagnation, the significance of inventions throughout history, the mindset of fearlessly refactoring code in Rust, the adoption of distributed systems, and fascination with 10 gigabit ethernet and concerns about fragile fiber optic cables.

Josh Clulow, Patrick Mooney, and Steve Klabnik join the hosts to discuss Helios, the OS that runs on the Oxide Rack. They touch on topics such as illumos, distros, Rust Tier 2 support, and Bryan's sleeper bug from 1991. They also explore the challenges of software distribution, developing software for Windows with Rust, memory usage, upstreaming code, and exploring computer architecture. The chapter descriptions include discussions on books, the Unix wars, the Helios sled, and database options.

Sean Klein, Rain Paharia, and Steve Klabnik join Bryan and Adam to discuss accelerating Rust builds, analyzing slow spots, cargo challenges, monomorphization, Buck 2 limitations, and Russ Analyzer improvements.

Simon Willison, a practical expert on open source large language models (LLMs), brings clarity to their workings and limitations. The podcast explores misconceptions surrounding open source AI, ethical concerns of open source training, and democratizing access to LLMs. It also discusses the impact of AI on various professions and the benefits of open source in search engines. The hosts express admiration for Simon and delve into the complexity of the web three and crypto space. Simon's work with Data Set, an open source tool for exploring and publishing data, is also discussed.

MIT Research Scientist, Michael Cafarella, joins Bryan and Adam for their predictions episode. Topics include past predictions, AI advancements, the iPhone adoption, Chach EPT and LUMOS commands, massive cuts in VMware, the future of internet search, recording meetings, and future CPUs.

George Cozma and Jordan Ranous join Adam and Bryan to discuss AMD's MI300 announcement and its implications to accelerated compute. They highlight the MI300A, which features CPU and GPU chiplets in the same package. The guests also discuss AMD's approach to hardware and software challenges, adoption of liquid cooling, challenges of training large models, and predictions for the future.

Nirav Patel, founder and CEO of Framework Computer, joins the hosts to talk about building a new repairable and open laptop company. They discuss Framework expansion cards, a gaming console from the community, and the Optima Braille laptop. They also touch on the history of talking computers and share their experiences with the Acer Ferrari 3400. The podcast covers topics such as the importance of long-term trust, financing challenges for hardware startups, the future of Framework Computer, and unlocking creativity with aging laptops.