エピソード

  • Ep. 18 — Anthropic's Identity Crisis & the Cache TTL Nobody Announced

    Ep. 18 — Anthropic's Identity Crisis & the Cache TTL Nobody Announced
    2026/04/13
    Anthropic's Identity Crisis & the Cache TTL Nobody Announced 00:48 — Anthropic's velocity-vs-quality moment — AMD says Claude regressed, community says stop shipping, HumanX says everyone's talking about Claude 02:17 — Anthropic quietly downgraded cache TTL on March 6th — developers found out through billing 03:47 — OpenAI silently removed Study Mode from ChatGPT 04:12 — Cirrus Labs to join OpenAI 04:36 — Trump officials may be encouraging banks to test Anthropic's Mythos model Daily AI News by Archer & Iris.
    続きを読む 一部表示
    7 分
  • Ep. 17 — Deep Dive: The context window arms race

    Ep. 17 — Deep Dive: The context window arms race
    2026/04/12
    Deep Dive: The context window arms race 00:48 — The context window arms race — deep dive Daily AI News by Archer & Iris.
    続きを読む 一部表示
    7 分
  • Ep. 16 — Deep Dive: The AI memory problem

    Ep. 16 — Deep Dive: The AI memory problem
    2026/04/11
    Deep Dive: The AI memory problem 00:48 — The AI memory problem — weights, context, external store, and the part nobody has solved Daily AI News by Archer & Iris.
    続きを読む 一部表示
    9 分
  • Ep. 15 — OpenAI's $100 Tier, Google's Gemini Goes 3D & The Industry Hits the Profit Cliff

    Ep. 15 — OpenAI's $100 Tier, Google's Gemini Goes 3D & The Industry Hits the Profit Cliff
    2026/04/10
    OpenAI's $100 Tier, Google's Gemini Goes 3D & The Industry Hits the Profit Cliff 00:48 — ChatGPT Pro now offers a $100/month tier targeting Codex power users 02:17 — Google's Gemini can now answer questions with interactive 3D models and simulations 03:47 — The AI industry's race for profits is now existential 05:17 — NYT op-ed: Anthropic's Mythos restraint is a terrifying warning sign 05:42 — Gemma 4 on llama.cpp is now stable 06:06 — Reverse engineering Gemini's SynthID detection 06:30 — Playing DOOM with 1.3M parameters: specialized small models vs LLMs 06:54 — CyberAgent scales AI adoption with ChatGPT Enterprise and Codex 07:18 — LLM Spirals of Delusion: auditing AI chatbot behavior over long conversations Daily AI News by Archer & Iris.
    続きを読む 一部表示
    9 分
  • Ep. 14 — Meta's Muse Spark, Claude Gets Agent Infrastructure & Mythos Under the Microscope

    Ep. 14 — Meta's Muse Spark, Claude Gets Agent Infrastructure & Mythos Under the Microscope
    2026/04/09
    Meta's Muse Spark, Claude Gets Agent Infrastructure & Mythos Under the Microscope 00:48 — Meta launches Muse Spark, its first model from Superintelligence Labs 02:17 — Anthropic launches Claude Managed Agents in public beta 03:47 — Claude Mythos system card draws 500+ comments as community dissects the benchmarks 05:17 — OpenAI releases Child Safety Blueprint for AI developers 05:42 — The vibes are off at OpenAI 06:06 — Developer resurrects a 30-year-old multiplayer game using Claude Code AI Current with Archer & Iris.
    続きを読む 一部表示
    8 分
  • Ep. 13 — Anthropic Hacks Everything & Sam Altman Can't Code

    Ep. 13 — Anthropic Hacks Everything & Sam Altman Can't Code
    2026/04/08
    Anthropic Hacks Everything & Sam Altman Can't Code 00:48 — A new Anthropic model found security problems ‘in every major operating system and web browser’ 02:17 — Google quietly launched an AI dictation app that works offline 03:47 — Intel will help build Elon Musk’s Terafab AI chip factory 05:17 — Testing suggests Google's AI Overviews tell millions of lies per hour 05:42 — Sam Altman's coworkers say he can barely code and misunderstands basic machine learning concepts Daily AI News by Archer & Iris.
    続きを読む 一部表示
    10 分
  • Ep. 12 — AI Takes Lab Control & Your Job's Hidden Data Point

    Ep. 12 — AI Takes Lab Control & Your Job's Hidden Data Point
    2026/04/07
    AI Takes Lab Control & Your Job's Hidden Data Point 00:48 — The one piece of data that could actually shed light on your job and AI 01:12 — Full-Duplex-Bench-v3: Benchmarking Tool Use for Full-Duplex Voice Agents Under Real-World Disfluency 01:36 — Comparative reversal learning reveals rigid adaptation in LLMs under non-stationary uncertainty 02:00 — Single-agent vs. Multi-agents for Automated Video Analysis of On-Screen Collaborative Learning Behaviors 02:23 — Toward Full Autonomous Laboratory Instrumentation Control with Large Language Models 02:47 — Steerable but Not Decodable: Function Vectors Operate Beyond the Logit Lens Daily AI News by Archer & Iris.
    続きを読む 一部表示
    11 分
  • Ep. 11 — Claude Users Pay Extra & Copilot Just for Fun

    Ep. 11 — Claude Users Pay Extra & Copilot Just for Fun
    2026/04/06
    Claude Users Pay Extra & Copilot Just for Fun — Daily AI News by Archer & Iris.
    続きを読む 一部表示
    11 分