Every Monday, 6am PT
The Intelligence Brief
The week's research on judgment, hiring infrastructure, and what AI is doing to both. Read, screened, and summarized by the team that has to act on it.
Week of November 16, 2026
In a randomized trial, 50 analysts followed an AI agent's hidden preference on 86% of choices while 15% credited it, eSkill's survey has 52% of HR professionals naming resumes the most exaggerated signal, and a 285-student study found approval-only AI workflows lowered quality, satisfaction, and diversity of work.
Read this week's briefInside this issue
- The Human Intelligence Edge
- The Broken System
- The SupaHuman Standard
- The Future of Work, Without the Clichés
Past issues
Analysts followed a hidden AI preference 86% of the time, and only 15% said so
Anthropic on hidden AI influence, eSkill on resume trust, and PNAS Nexus on where human agency belongs in the workflow.
An application can now be scored on which model wrote it, and AI explanations without feedback made teams worse
The audit where the drafting model moved the score, the resume summaries that stayed factual and changed tone, and the condition that decides whether pairing helps.
AI screeners favored their own rewrites by 23 to 60%, and assisted consultants kept no skill after the tool left
What happens when the screener and the candidate run the same model, the consultants who performed better and learned nothing, and the 420 study synthesis on where pairing loses.
Thirty seconds more per resume moved an AI-skipped candidate 3 to 4%, and assisted output keeps overstating solo skill
What 30 extra seconds of review does for a candidate the AI skipped, two labs finding assisted gains that vanish in solo work, and the interaction pattern that keeps ideas and ownership with the person.
71% say AI makes critical thinking more important, and the heaviest users trust their own reasoning least
SHRM on the 71% who say AI raised the value of critical thinking, what 1,923 heavy GenAI users report about their own reasoning, and the validity gap between a structured interview and a conversation.
80% say AI made them more productive, 6% of their organizations qualify as high performers
McKinsey on the 80/6 gap between personal productivity and company performance, IBM's CHROs ranking the ability to override AI first, and what 30,646 participants say about interview scorecards.
Adaptability demand went up fivefold, and the hours AI saves come back as verification
MGI on adaptability demand up fivefold, the seven hours a week AI gives scientists back and where they go, and what a shared screening algorithm does to a pipeline.
De-skilling is being observed now, and referrals are coming back
BCG on de-skilling leaders already observe, the deference rate behind negative synergy, WEF on the return of the referral.
Deference is the failure mode, and almost nobody verifies
WSJ Intelligence on the verification gap, a 535-person study on negative synergy, Gartner on AI-written applications.
Judgment is the premium, and adoption is shallower than it looks
WSJ Intelligence on the skills companies must protect, Deloitte on judgment as the top behavior, NBER on adoption that is wide but shallow.
Judgment, de-skilling, and the 19% gap
Harvard on evaluator judgment, Stanford on young-worker employment, BCG on organizational de-skilling.