New Routledge research monograph

Dark Sides of Leadership in the AI Era

Opening manuscript: front matter, Preface, and Chapter 1. Read it in the scrollable frame below. The working paper, accepted Boje–Rosile chapter, and research tools remain available further down the page.

Book summary

Dark Sides of Leadership in the AI Era asks a question that is often obscured by promises of efficiency, innovation, and artificial superintelligence: what kinds of leadership are being written into the systems now reorganizing work, public life, and the conditions of the planet? This Routledge research monograph treats AI not as a disembodied tool and not as a single leader’s private intention, but as a layered organizational accomplishment. Executive speeches, product design, alignment policies, cloud infrastructures, consulting practices, investor stories, data work, water and energy demands, and the everyday conduct of users all enter the story.

The book develops Ghost Vortex as a Boje proposed construct for studying what returns beneath official accounts of responsible AI. The term joins the ghost—what has been declared absent but persists—with the vortex, a patterned movement that gathers dispersed traces. Ghost Vortex directs attention to the palimpsest of AI: answers that appear new may carry earlier inscriptions of corporate rhetoric, hidden labor, procurement decisions, guardrails, training data, and material arrangements. Plato’s cave remains important here. We meet the answer on the screen as a shadow and must ask what relations, exclusions, and costs make that answer possible.

The plural Dark Sides matters. This is not a book about declaring leaders villains, nor about pretending that a model has a single author. It is a research program for tracing interacting sides: moral answerability and evasion; leadership discourse and product behavior; consent and contestation; worker experience and automated management; data-center growth and community consequences; promises of safety and the practical capacity to correct harm. The book brings antenarrative, Bakhtinian answerability, qualimetrics, and SEAM—the socio-economic, environmental, and material hidden-cost ledger—into conversation with the emerging AI industry.

Its method is deliberately comparative and open to challenge. It collects public leadership discourse, studies AI products through repeatable prompt protocols, and reads the resulting transcripts alongside evidence of infrastructure, labor, governance, and affected voices. Seven-point coding makes patterns discussable; it does not replace interpretation, witness, or historical context. The goal is not to manufacture a scorecard that closes inquiry. It is to make visible the questions that official stories leave beneath the surface: Who benefits? Who bears the hidden cost? Who can contest an AI-mediated decision? Who is answerable when the story breaks?

The manuscript is developing through public dialogue. October Wednesday sessions invite readers, researchers, students, practitioners, prospective authors, and critics to read with the work rather than only receive it. The meetings are held October 7, 14, 21, and 28 at 9:00 a.m. Pacific, 10:00 Mountain, and noon Eastern. Participants may register through Eventbrite or request a Zoom calendar invitation from David through the World Scientific contact page. The invitation is simple: let there be light on the shadows, and let the people living with AI enter the story.

Opening manuscript for reviewer dialogue

David M. Boje · Routledge research monograph in development
Open full reader
Dark Sides of Leadership in the AI Era by David M. Boje Routledge research monograph
Research, dialogue, and public tools

Let there be light on the shadows.

Dark Sides of Leadership in the AI Era studies Ghost Vortex, palimpsest, organizational storytelling, infrastructure, and moral answerability where AI systems meet workers, communities, institutions, and leaders.

Visitor number: 00001
Ghost Vortex illustration showing answerability, hidden labor, infrastructure, and trusted advisor models
Penn State webinarSeptember 25, 2026 · What Is Storytelling Practice in OD and C?World Scientific LabSeptember 30 launch · then October WednesdaysAI Complicity CheckerTrace data, decision, labor, material, and answerability entanglementsAI company mapsAI firms, consulting firms, ERP platforms, and Walmart Sparky

Public conversations

Events that put the book into dialogue.

The book develops through public questions rather than behind a closed wall: one Penn State professional-development webinar and the World Scientific Series 3 working laboratory.

Friday · September 25, 2026

What Is Storytelling Practice in OD and C?

Penn State WFED/OD and C Professional Development Network · 60 minutes

9:00 a.m. Pacific / 10:00 a.m. Mountain / noon Eastern. A conversation for external consultants, HRM and HRD practitioners, internal OD, educators, researchers, and higher education confronting the AI transformation.

Register for the Penn State webinar
Wednesday launch · September 30, 2026

World Scientific Series 3 Working Lab

Business Storytelling in the Age of AI · 60 minutes

9:00 a.m. Pacific / 10:00 a.m. Mountain / noon Eastern. New editors present developing volume ideas and invite prospective editors and authors. October sessions continue October 7, 14, 21, and 28. Visitors may request a Wednesday invitation and David can add them to the calendar invitation.

Open working text

Read here. Do not download first.

The working paper and the accepted Boje–Rosile chapter appear in scrollable readers. The PDF files must sit in this same Darkside folder for the embedded readers to work.

Palimpsest and Ghost Vortex in the Dark Sides of AI Leadership

Research working paper for the Routledge monograph
Open PDF in a new tab

Dark Side of Leadership Storytelling

David M. Boje and Grace Ann Rosile · accepted chapter
Open PDF in a new tab

Interactive inquiry

Find Out Your AI Algorithm Entanglements

Start with the technologies you use, buy, sell, or work beside. Then check the forms of entanglement you recognize. The tool returns five priority research flags, not accusations or determinations of wrongdoing. Its purpose is to show the overlapping AI-provider, consulting, enterprise-platform, labor, and infrastructure layers that deserve closer inquiry.

Check every relation that applies

Research directory

Three corporate layers and a retail case.

These are initial ten-company research maps, not a revenue league table or a claim that any individual “is” a product. Each entry names a leader, a prominent product or platform, and the research layer to follow. Verify current roles and products against primary sources when a chapter is written.

Layer one

AI firms and products

FirmLeaderProduct under study
NVIDIAJensen HuangGPU and CUDA infrastructure
OpenAISam AltmanChatGPT
AnthropicDario AmodeiClaude
xAIElon MuskGrok
MetaMark ZuckerbergMeta AI and Llama
MicrosoftSatya NadellaCopilot
Alphabet GoogleSundar PichaiGemini and AI Overviews
AmazonAndy JassyAlexa Plus and Bedrock
OracleLarry EllisonOCI Generative AI
DeepSeekLiang WenfengDeepSeek models
Layer two

Consulting and guardrail firms

FirmLeaderAI practice or platform
AccentureJulie SweetAI Refinery
DeloitteJoe UcuzogluAI and data consulting
PwCMohamed KandeAI transformation
EYJanet TruncaleEY.ai
KPMGPaul KnoppAI and trusted AI services
McKinseyBob SternfelsQuantumBlack and Lilli
BCGChristoph SchweizerBCG X
BainChristophe De VusserAI consulting
CapgeminiAiman EzzatCapgemini Invent AI
IBM ConsultingArvind Krishnawatsonx
Layer three

Enterprise platforms and ERP

FirmLeaderPlatform or product
SAPChristian KleinJoule
OracleLarry EllisonFusion Applications AI
SalesforceMarc BenioffAgentforce
MicrosoftSatya NadellaDynamics 365 Copilot
WorkdayCarl EschenbachWorkday AI agents
ServiceNowBill McDermottNow Assist
IBMArvind Krishnawatsonx
InforKevin SamuelsonInfor GenAI
SageSteve HareSage Copilot
NetSuiteEvan GoldbergNetSuite AI
Retail case

Walmart and Sparky

Walmart describes Sparky as its generative-AI shopping assistant in the Walmart app. It is a useful retail case for the book because it moves from recommendation into the practical conduct of shopping: search, comparison, basket building, and purchase.

The research question is not whether convenience is bad. It is what changes when a product recommendation becomes an agentic retail relation: whose data, product claims, seller interests, labor conditions, supply-chain effects, and correction paths are visible to the customer?

Read Walmart's Sparky announcement

Book terms

A glossary that opens rather than closes the conversation.

These are working definitions for the monograph. Terms introduced by Boje are marked as proposed constructs; their purpose is to make visible what conventional leadership and technology vocabulary leaves beneath the surface.

Ghost Vortex — Boje proposed construct

Ghost evokes the return of a prior presence that official speech declares absent; vortex, from Latin vortex, names a whirling current that draws dispersed matter into a patterned movement. Ghost Vortex is Boje’s proposed term for the recursive reappearance of buried organizational assumptions in AI-mediated conduct. It does not claim that a leader’s private personality is literally inside a machine. Rather, it directs analysis to the traceable layering of executive rhetoric, governance, training and alignment choices, consulting guardrails, client policy, retrieval systems, labor arrangements, and material infrastructure that shape what an AI system can name, defer, or render unsayable.

Palimpsest

From Greek palimpsestos, “scraped again,” a palimpsest is a parchment or other writing surface reused after an earlier inscription has been erased incompletely. In manuscript studies, recovery of the undertext makes visible the material history of reuse. Boje employs palimpsest as an analytic discipline, not a decorative metaphor: the apparent freshness of an AI answer may overlay prior institutional inscriptions, while residues remain available for recovery through transcripts, product policies, procurement records, labor accounts, and community testimony.

Antenarrative

Boje coined antenarrative by bringing together ante, Latin for “before” or “in advance of,” and narrative. It concerns the fragmented, retrospective, prospective, and contested story fragments that precede and exceed the stabilized plot. Antenarrative is therefore not an unfinished narrative waiting to become coherent. It is a field of bets among multiple possible futures and meanings. In this book, it prevents the company statement, safety policy, or model response from closing inquiry before affected people can enter the story.

Bakhtinian answerability

Answerability translates Bakhtin’s concern with the situated, once-occurrent act for which a person cannot evade responsibility by hiding behind a general rule. It is stronger than consultation, transparency, or an explanation of process. An answerable organization identifies who can respond to an affected person, what evidence will be considered, what correction or redress is possible, and who remains responsible when an AI system causes a harmful or exclusionary outcome.

Antenarrative Shutdown — Boje proposed construct

Shutdown names a foreclosure rather than an ordinary difference of opinion. Antenarrative Shutdown is proposed for a documented and reproducible pattern in which an AI system cannot or will not sustain a relevant account containing multiple affected voices, while it remains capable of responding to adjacent, less disruptive prompts. The construct must not be used for a single outage, a routine safety refusal, or an inconvenient answer. It requires dated transcripts, comparable prompts, and a careful distinction between technical instability and patterned narrative closure.

Tesseract and the XYZA dimensions

A tesseract is the four-dimensional analogue of a cube; the term provides an image for relations that cannot be flattened into a single organizational timeline. Boje’s XYZA Tesseract brings together X, task; Y, humanistic; Z, becoming; and A, answerability. Task asks what the AI arrangement accomplishes; humanistic asks whose dignity, capability, and voice are at stake; becoming attends to emergence, future possibility, and transformation; answerability asks who must respond for consequences. The dimensions are held in tension, not used as a mechanical scorecard.

Qualimetrics

Qualimetrics joins qualitative interpretation with explicit metrics. In the socio-economic management tradition, it is used to make hidden costs, organizational processes, and human potential discussable without pretending that every important reality is reducible to a number. In this study, ratings can make patterns across dated AI sessions comparable, but they remain subordinate to the full transcript, coding rationale, historical context, and the possibility that an affected witness changes the interpretation.

AI algorithm entanglement

Entanglement names relation without separability: an organizational AI practice cannot be understood only as a model, a prompt, or a decision made by a user. AI algorithm entanglement is the web of data, cloud infrastructure, vendor contracts, workplace routines, labor, local water and energy demands, and governance arrangements through which an organization uses or sells AI. The complicity checker asks where that web is visible, who can challenge it, and how a decision might be repaired.

Restorying

Restorying is the practice of revisiting a settled account so that excluded events, voices, relations, and possible futures can re-enter. It is not public-relations reframing. In Boje’s use, restorying is ethical and practical work: it opens a story that has become totalizing, lets people locate their own participation and capacity, and makes different organizational action imaginable.

AI Bets on Extinction illustration

Plato's cave and the Ghost Vortex

The answer on the wall is never the whole event.

Plato’s cave offers a beginning image for the research: people encounter shadows and struggle to infer the arrangements that cast them. Ghost Vortex adds the palimpsest. The screen is not only a wall for projected shadow; it is a surface written over by foundation models, provider alignment, consulting guardrails, client policies, retrieval systems, access controls, and human labor.

The inquiry is therefore double: trace the shadows back to their material and organizational conditions, and ask who can interrupt the story that presents those conditions as inevitable.

Explore Ghost Labor at Storying.site

Start with where you shop or work.

Select the settings that are part of your everyday life. These selections identify contexts for research; they do not say that a named company used a particular AI, surveillance, pricing, or employment practice.

Public research and regulatory activity identify possible risks in retail personalization and pricing, travel booking, worker monitoring, automated employment decisions, and algorithmic management. The question is what is documented in your specific setting.

Opening manuscript for reviewer dialogue

Front matter, Preface, and Chapter 1 · David M. Boje · Routledge research monograph in development
Open full reviewer reader
Earlier working papers and accepted chapter

These supporting texts remain available for context. The new book manuscript above is the primary reviewer text.