The OS powering Assistantly
All issues

Intelligence Brief

Week of
October 5, 2026

Your weekly edge, the research, frameworks, and signal that separate operators from spectators. Read and summarized so you do not have to.

Published Monday, October 5 at 6:00 AM PT. 12 sources, 7 minute read.

01

The Human Intelligence Edge

What makes humans irreplaceable in the AI era.

NatureSeptember 23, 2026

How to stay smart in the age of AI: the science of critical thinking

Helen Pearson

Why it matters. Pearson gathers what is actually known about AI and independent reasoning, and the headline numbers are beliefs rather than measurements: 90% of more than 1,000 surveyed US faculty expect generative AI to reduce students' critical thinking, and about one in five AI-using students in a separate survey say working through a problem without it has become harder. Self-reports are weak evidence of cause and strong evidence of experience. The students noticing the slide are telling you the skill is maintained by practice rather than possessed, which is the whole argument for making a candidate reason in front of you at least once, unassisted.

Read the source
Boston Consulting GroupSeptember 24, 2026

How Leaders Can Preserve Human Advantage in the AI Era

James Tucker

Why it matters. Tucker's mechanism is specific: performance holds up when people critically evaluate what the model returns, and persuasive output is the kind accepted fastest. Set it beside the Nature feature above and the two halves line up. Executives report the organizational symptom, students report the individual one, and nobody in either group decided to stop thinking. The work simply stopped requiring them to do it out loud. That is a problem in how roles are written before it is a problem in a person, and it is cheaper to fix in the job description than in the hire.

Read the source
World Economic ForumSeptember 21, 2026

With AI becoming more capable, here's why people must too

Shuvasish Sharma and the Global Future Council on Jobs and Frontier Technologies

Why it matters. Analytical thinking still matters to seven in ten employers, while the 2025 study of 319 knowledge workers WEF cites found higher confidence in generative AI tracking with lower self-reported critical-thinking effort. The two facts only look contradictory. Demand for the skill is climbing at the same time the daily practice of it is being handed off, which is how a shortage forms without anyone announcing it. What is left for people is verification, integration, and stewardship, so hire the person already doing all three before anyone asks them to.

Read the source
02

The Broken System

Why resumes, interviews, and hiring infrastructure are failing.

MIT NewsSeptember 29, 2026

The effects of an algorithmic monoculture depend on the details

Adam Zewe, reporting on research by Brian Hedden and Manish Raghavan

Why it matters. Hedden and Raghavan model what happens when many firms screen candidates with the same algorithm, and the finding is that the details decide it. Shared models are not automatically worse than scattered human judgment, and in some setups they are better. What changes the answer is candidate agency, recourse, and how the decision is structured. The practical version for a hiring team: ask what a rejected candidate can do next, whether a second look is possible, and whether the screen you run is the one your competitors run. If it is, a single bad read does not cost someone one job, it closes the whole market at once.

Read the source
SHRMSeptember 15, 2026

Fraudulent Candidates at Scale: Are AI-Generated Applications Breaking Recruitment?

SHRM Advisor, citing Gartner survey data

Why it matters. Gartner found 39% of 3,290 candidates used AI somewhere in the application, and 6% of a separate 3,000 admitted to interview fraud. Those are two different problems that keep getting treated as one. Widespread assistance makes polish uninformative, which is solved by testing the work instead of the artifact. A smaller rate of outright fraud makes identity something you have to establish, which is solved once, at the door, before anyone spends an hour in an interview. Conflate them and you end up policing writing style while the harder failure walks straight through.

Read the source
arXiv preprintSeptember 17, 2026

When Hiring Becomes Agent-Mediated: Evaluating Access and Recurrence in Two-Agent Resume Screening

Jian Gao and Hang Jiang, accepted to the REALM Workshop at EMNLP 2026

Why it matters. Across 600 constructed resume and job pairs, a two-agent screen that exchanged evidence advanced a different set of candidates than a single scoring call, lifting pass rates in some model runs. The configuration is the policy. Teams install an automated first pass as though it were a measuring instrument, when what they have installed is a position on which evidence counts and how many times it gets reconsidered. Read your own setup back before you defend one of its rejections to a candidate.

Read the source
03

The SupaHuman Standard

The behavioral science of who actually thrives.

McKinsey Global InstituteSeptember 29, 2026

Workforce in motion: Skills and pathways to future jobs in the United States

María Jesús Ramírez, Kweilin Ellingrud, Tanguy Catlin, Diego Castresana, and Anna Kortis

Why it matters. Demand for adaptability in US job postings rose about fivefold between 2022 and 2026, with resilience, curiosity, and willingness to learn up roughly threefold. A posting is a record of what an employer will commit to in writing, so the slope is the interesting part: these moved from culture-deck language to stated requirements inside four years. They are also the only requirements on a posting that say anything about what someone can do a year out, once the tools and the role have both moved. A credential describes a career in the past tense.

Read the source
World Economic ForumSeptember 16, 2026

What skills do employers want from young people in the AI age?

Charlotte Edmond, citing WEF and PwC analysis

Why it matters. PwC analysis cited by WEF puts the most AI-exposed junior roles at seven times more likely than the least-exposed to ask for traditionally senior skills, leadership among them. Entry level now opens where mid level used to sit. It is a labor-market pattern rather than a test of individual performance, and it still changes what an entry-level interview is for. You are no longer checking whether someone can run a defined task. You are checking what they do when the brief moves under them in week two.

Read the source
WSJ IntelligenceSeptember 10, 2026

New WSJ Intelligence Survey Highlights the Human Skills Companies Must Protect as AI Enters the Workplace

WSJ Intelligence for Philip Morris International, via Business Wire

Why it matters. Across more than 2,500 professionals in five countries, moral judgment and ethics (42%) and empathy and trust-building (40%) came out both most important and least replicable. This is employer sentiment, not performance data, and it is still the clearest statement of what companies believe they cannot buy from a vendor. Neither attribute reaches a resume, and neither survives an interview answer rehearsed the night before. You learn it by handing someone a decision with a cost attached and watching how far they carry it.

Read the source
04

The Future of Work, Without the Clichés

Where AI and human collaboration is actually heading.

MIT FutureTechSeptember 16, 2026

AI in Science: Early Insights

Mihai Codreanu, Alex Imas, Juan Mateos-Garcia, et al.

Why it matters. Triangulating 15 million Gemini interactions, more than 2,600 specialized models, and a survey of more than 600 scientists, the study finds researchers saving close to seven hours a week while reporting heavier verification demands and a growing backlog of hypotheses they cannot get through. The time did not leave the system, it moved. Hours returned at the production end arrive downstream as checking, prioritizing, and deciding what is worth pursuing, which is judgment work and the hardest kind to staff. Hire against the bottleneck you are about to have, not the one that just cleared.

Read the source
Boston Consulting GroupSeptember 24, 2026

Five Ways That AI Front-Runners Change How Work Gets Done

Sagar Goel, Julie Bedard, Matthew Kropp, Ashley Sim, and Charikleia Kaffe

Why it matters. A qualitative study across 50 companies finds the front-runners redesigning work around outcomes: AI produces, people direct and evaluate, and a team owns the result. What separates them is not model access, it is decision rights. Who approves, who answers for a wrong output, and what the workflow does with a disagreement. Those questions have organizational answers and they get settled before any tool arrives. A company that has not answered them buys the same model as everyone else and gets less out of it.

Read the source
arXiv preprintSeptember 15, 2026

Available but Unclaimed: An Empirical Study of Human-AI Synergy

Robin Welsch, Michelle Rausch, Daniela Fernandes, et al.

Why it matters. In a 535-person reasoning study, assistance beat unaided work while the human and AI pairing did not reliably beat the assistant on its own, and participants followed the advice at times when it was wrong. A human in the loop is a seat, not a safeguard. What converts one into the other is calibration: knowing which claims to check, which to push back on, and which to let stand. An hour of observed work shows you that. No credential a candidate can send you does.

Read the source
Previous issueWeek of September 28, 2026 All issuesBrowse the archive