Research

Seven years. Two hundred organisations. One question.

Since 2019 I have been asking organisations across thirty countries the same thing: are you using AI, or has it started using you? That fieldwork, through interviews, advisory engagements and surveys with leaders in banking, logistics, media, law, professional services, technology and education, is where the SuperSkills framework comes from.

Here for the research on AI and human capability? Start with the evidence, or go straight to the full index of every page.

It is not the same thing as the evidence. This site separately publishes 312 graded studies, each with a line on what it does not prove, and 780 questions marked answered, partial or open. And neither is the same as the argument: the positions I take are set out on their own page, so you can see where the evidence stops and I start. Those three are kept apart deliberately. Most writing on AI runs them together.

The Drift Index

An annual reading of organisational judgement.

Everything below this line is synthesis: what the published evidence establishes, how strong it is, and what it does not support. Read together it is also an argument for something that does not exist yet. No validated instrument measures whether an organisation is losing capability, which is why this research lists capability debt as a framework rather than a finding. The Drift Index is the attempt to close that gap: an annual reading of organisational judgement, where it is holding and where it has moved without anyone deciding. First edition December 2026.

The first edition is compiling now. If you would like your organisation included, get in touch. Until it publishes, the honest position is the one set out in what we actually know: the mechanism is well evidenced, the organisational measurement is not.

The drift versus design matrix A two-by-two matrix. Vertical axis: agency, low to high. Horizontal axis: awareness, low to high. Low agency and low awareness: the Sleepwalkers, who do not notice the drift. Low agency and high awareness: the Stuck, who see the patterns but cannot break them. High agency and low awareness: the Programmed, who optimise without questioning. High agency and high awareness: the Designers, who see the systems and shape them. AGENCY High Low AWARENESS Low High The Sleepwalkers They do not noticethe drift. The Stuck They see the patternsbut cannot break them. The Programmed They optimisewithout questioning. The Designers They see the systemsand shape them.
The drift versus design matrix. Where an organisation sits depends on two things: whether it can act on what shapes its choices, and whether it can see those forces at all. From SuperSkills (Kogan Page, 2026).
The work

Six ways in.

The map of the questions, what the evidence actually supports, the chronology of the writing that moved the argument, the position this research will defend, the board version, and the tool. Everything else is in the full index below.

The questions

The public map of the territory. 780 real questions about what increasingly capable AI does to humans, grouped into 18 areas and marked answered, partial or open. The unanswered ones stay visible on purpose.

Enter →

What We Actually Know

The state of the evidence. Nineteen claims separated into strong evidence, emerging evidence, and what remains unknown, each graded and linked to its primary source. Including the questions this research cannot yet answer. Reviewed quarterly.

Enter →

The Best Writing on AI

A chronology from March 2023 to August 2026, because the order is the argument. What actually moved the conversation, why the founding estimates are still quoted over the measurements that complicate them, and how the question changed underneath everyone.

Enter →

Human in the Loop Is Not a Safeguard

A position page. 370 effect sizes across 106 experiments say the arrangement most organisations have installed makes results worse, not better. With the strongest objections given in their strongest form, and what would change the position stated rather than withheld.

Enter →

What a Board Should Ask

Twelve questions, each paired with the answer that should worry you. Not an agenda item: ask two or three in passing and listen for hesitation. The twelfth does more work than the other eleven combined.

Enter →

The Delegation Boundary Map

"Keep a human in the loop" does not say which human, at which point, or on what grounds. Nine stages of work, and for each: what the human keeps, what the machine may do, who verifies, and who is accountable. With a downloadable working grid.

Enter →

Intelligence

Counted, and checkable

312 graded studies, each with what it does not prove. 780 questions, mapped and answered or marked open.

The argument

Drift versus Design, and the concepts under it

Book research

What sits beneath SuperSkills

Writing

Essays, most Sundays since 2017
A woman crossing a city street at dusk, eyes closed and arms outstretched, following lit arrows painted on the road

The Architecture of Drift

How AI became the most sophisticated drift machine ever built, and how to use it without being used by it.

Read → An airliner flight deck at dawn, instruments lit, the aircraft flying itself

Are you flying, or being flown?

Algorithmic drift: the slow process by which systems start making your choices so smoothly you stop noticing.

Read → An open drawer of tangled cables, adapters and discarded instruction leaflets

We’ve been the AI all along

What changes when millions of people can build any software they can imagine.

Read → An archivist seated on the floor of a manuscript room by lamplight, reaching for a bound volume

The Crossing

Hope is not a feeling. It is a skill, and it can be practised.

Read → The words Synthetic Seniority set in heavy type, the letters misprinted and slipping out of register

Solving Synthetic Seniority

Why AI can make people look senior without the reps that make them senior, and what to do about it.

Read → The Box of Amazing logo

The full archive

Essays and curation since 2017, a new one every Sunday. Free.

Browse all →

Where these essays became research: The Architecture of Drift → Drift versus Design, Solving Synthetic Seniority → Synthetic Seniority, We’ve been the AI all along → Staying Valuable.

Bylines

Published elsewhere
An illustration of a hand holding a pair of balanced scalesNew York Observer · Jul 2026

The future of AI: power, judgment and trust

On drift, design and staying the author of your own judgement as AI takes over more decisions.

Read → Two people silhouetted against a wall of illuminated AI dashboards and chartsCEOWORLD Magazine · Jul 2026

Drift versus design

Why most companies mistake activity for transformation, and how to design change on purpose.

Read → Two people shaking hands over an open notebook and a financial reportThe Bookseller · 2026

Artificial intelligence: the chain of trust

What publishing, and every knowledge industry, stands to protect or lose as AI reshapes trust.

Read → A close-up of a pick-and-mix of brightly coloured boiled sweetsIrish Tech News · 2026

You’re not adopting AI. You’re paying for it.

Why most AI transformation spend fails to change how people actually work.

Read → A hand touching a glowing network of illuminated person iconsEntrepreneur · 2026

AI amplifies human decisions, it doesn’t go rogue

The real risk is not runaway machines. It is the human judgement we hand over without noticing.

Read → A presenter addressing a group seated around a boardroom table with laptops openTechRound · 2026

Are universities teaching the skills employers actually need?

World Youth Skills Day 2026: the widening gap between what is taught and what the AI-era workplace demands.

Read → The European Business Review cover image for Leaders in the AI Age, a group around a table marked Artificial IntelligenceThe European Business Review · 2026

The real AI risk is accountability, not automation

The gaps open in leadership decisions, not in the machines. What accountable AI adoption actually requires.

Read → A robotic hand and a human hand fitting two jigsaw pieces togetherWorldatWork · 2026

From skills management to capability design

Why HR’s next job is designing capability, not managing skills, as AI resets what expertise means.

Read →

Everything, in one place

The full index.

All 234 research pages, grouped into 14 sections, with a note on what each one is for. Generated from the estate rather than maintained by hand, so nothing can be published and then be impossible to find.

Open the full index →

Start with a question

The questions this research answers.

The hubs below are organised by idea. This is the same body of work organised by the problem you arrived with. Each links to the main answer.

Does AI weaken human judgement?Not on its own. It reduces the demand for judgement, and demand is what builds it. Does AI weaken critical thinking?It does not disappear. It migrates to verification, and thins unless you design against it. What human skills become more valuable as AI improves?Not the ones on the reassuring list. The ones that raise the expertise of the work that remains. What happens when AI removes junior work?The wrong worry is graduate unemployment. The right one is where senior people come from. How should humans and AI make decisions together?Not by adding a person at the end. Across 106 studies, undesigned pairing underperformed both. How can organisations prevent AI-driven deskilling?By redesigning how people learn at the same pace as they redesign how work gets done. How do I use AI without becoming dependent on it?Form the view before you prompt, and keep one unaided rep of whatever carries your value. What should leaders keep human?Accountability, recognition, and the origination of the question. The task list will keep moving.

All 780 questions about humans and AI →

Questions about the research

How the work is done.

How was the SuperSkills research conducted?

Through interviews, advisory engagements and surveys with leaders across more than 200 organisations in over 30 countries, from 2019 to 2026. The sectors span banking, logistics, media, law, professional services, technology and education. The research base is where the seven SuperSkills framework comes from.

What is the Drift Index?

The Drift Index is an annual reading of organisational judgement: where it is holding, and where it has moved without anyone deciding. It draws one senior respondent per organisation and publishes each December, with the first edition in December 2026.

Can my organisation take part in the Drift Index?

Yes. The first edition is compiling now, with one senior respondent per organisation. Get in touch through the contact page and mention the Index.

Is the research peer-reviewed?

No, and it does not claim to be. It is practitioner research: systematic, longitudinal and cross-sector, conducted through direct engagement rather than academic publication. Where the book cites external studies, those citations go to the primary sources.

Can I use Rahim's frameworks in my own work?

Yes, with attribution. The drift versus design matrix, synthetic seniority and the other frameworks are published and citable: name the source and link to the original. For commercial licensing, training use or reproduction in publications, get in touch.

How this research works, in full →

"The goal isn't more technology. It's more capable humans."
Rahim Hirji
Box of Amazing

New thinking, every Sunday.

A weekly essay and curation on AI, human behaviour and the future of work, written every week since 2017 and read by more than 25,000 leaders. Prefer less often? Choose the monthly roundup instead. Free either way.

Or open Box of Amazing →

Put it to work

The ideas are the start. The value is in the room.

Reading is the backup. The change happens when Rahim is on your stage or working with your leaders. Bring this work into your organisation through a keynote or an advisory engagement.

Keynotes Advisory & coaching Book a call