• Focus Paper: AutoAnalysis: An automated, self-updating meta-analysis of human-AI collaboration
    Nov 18 2026

    A single-paper review of AutoAnalysis, a preprint from researchers at Harvard Business School and MIT Sloan that uses AI agents to run a continuously updating meta-analysis of human-AI collaboration. Across 1,955 effect sizes from 420 papers, people working with generative AI outperformed people working alone, but the pairing fell short of the better of the human or the AI alone. We cover how the pipeline works, what seems to help, such as division of labor and AI explanations, and the limits of an AI-run review.

    AI-narrated review of a single paper. The paper is named in the episode.

    Show More Show Less
    11 mins
  • Issue 25: Human by Exception
    Nov 16 2026

    This week's research review looks at supervision spreading thin: two combat drones flown under one operator, a targeting system described as autonomous by default and human by exception, and new fighter AI agents. Lab studies show operators working fastest with clear displays over opaque AI recommenders, trust spreading between teammates under time pressure, and people following AI advice that is wrong when they get no feedback. Also covered: stand-down tests for AI agent teams and the autonomous weapons talks opening in Geneva.

    AI-narrated. Sources are named in the episode.

    Show More Show Less
    19 mins
  • Issue 24: Is the Human in the Loop Real?
    Nov 9 2026

    The debate shifts from whether humans stay in control to how control is delivered: a cockpit tablet, a target cue, an approval step, and approval steps an AI agent can route around.

    AI-narrated. Sources are named in the episode.

    Show More Show Less
    19 mins
  • Issue 22: Do Humans and AI Make Each Other Better?
    Oct 26 2026

    The military pushes AI agents toward real operations, including cyber agents with minimal oversight, while researchers get stricter about whether human-AI teams outperform either one alone.

    AI-narrated. Sources are named in the episode.

    Show More Show Less
    12 mins
  • Issue 21: Oversight as a Job
    Oct 19 2026

    Autonomy scales from single robots to whole formations. Researchers ask what oversight looks like as an actual job, with hours, fatigue and a screen that may or may not help.

    AI-narrated. Sources are named in the episode.

    Show More Show Less
    13 mins
  • Issue 19: Autonomy Becomes an Institution
    Oct 5 2026

    Human-autonomy teaming moves out of the lab and into institutions: new commands, fielding deadlines and job codes. The research question underneath it all is how much one person can review before oversight breaks down.

    AI-narrated. Sources are named in the episode.

    Show More Show Less
    13 mins