Across the seven previous instalments of this series, a single argument has been gradually assembled. Generative systems do not so much disrupt the content of work as alter the economics of the signals through which organisations infer competence, allocate authority and decide whose contribution counts. The present essay closes the series by following that argument to its most consequential implication, which concerns the population of people whose cognitive style has, for most of the modern industrial era, sat awkwardly inside organisational frames built for someone else. The shift in question is neither inclusive nor exclusionary by intent. It is structural, it is already underway, and it has begun to reorder the relative economic value of cognitive styles in a manner that the management literature is only just beginning to register.
The argument that follows is intended as a capstone, and it deliberately resists the temptation to repeat material already developed in earlier essays. Where prior pieces traced the surface of the change, this one attends to the deeper question, namely what an honest reading of the available evidence suggests about the position of neurodivergent professionals inside organisations five years from now, and what executives should be doing about it before the answer becomes self-evident in the form of attrition statistics, missed product opportunities and a quiet realisation that the firm has been losing access to a category of cognition it never knew it depended on.
The shift in question is neither inclusive nor exclusionary by intent. It is structural, it is already underway, and it has begun to reorder the relative economic value of cognitive styles in a manner that the management literature is only just beginning to register.
A Population the Modern Firm Was Not Designed Around
The modern firm, in the form that emerged from the second industrial revolution and was refined by Taylor, Fayol and the early American management schools, was designed around a particular profile of cognition. It assumed sequential processing, predictable attention, fluent verbal exchange in standardised settings, calibrated emotional display, and the capacity to translate internal reasoning into the formats expected by colleagues, superiors and external stakeholders. These assumptions were not articulated as a profile of any particular kind of mind. They were embedded in the architecture itself, in the design of meetings, in the rhythm of performance reviews, in the layout of open offices and in the cadence of the working day. As an earlier essay in this series argued, what looked like a neutral organisational template was in fact a cognitive filter, and one that admitted some kinds of mind more readily than others.
The proportion of the working-age population that sits some distance from that template is considerably larger than most boards assume. Estimates from Deloitte place the share of the global population with some form of neurodivergence at between fifteen and twenty per cent, a figure replicated by SAP, the World Economic Forum and the EY Global Neuroinclusion at Work Study 2025.1 The same EY study, conducted across roughly four thousand five hundred respondents in seventeen countries and published in July 2025, finds that neurodivergent employees report disproportionately strong capabilities in problem solving, creative pattern recognition, hyperfocus on tasks of genuine interest and sustained analytical depth. It also finds that the same employees report disproportionately negative experiences with performance evaluation, recruitment processes, open-plan environments and the ambient social load of contemporary office life.2 The asymmetry is precisely what one would expect of a population whose cognitive endowment is misaligned with the surface conventions of the firm without being misaligned with the underlying work.
Robert Austin and Gary Pisano argued in their 2017 Harvard Business Review essay, which has become the canonical reference in this area, that this misalignment represents not a problem of accommodation but a forgone source of competitive advantage. Their analysis of the early neurodiversity hiring programmes at SAP, Hewlett Packard Enterprise, Microsoft, Willis Towers Watson, Ford and Caterpillar identified a recurring pattern, in which firms that redesigned the entry point to the organisation in order to admit a different kind of mind found themselves with capabilities they had previously been unable to recruit at any price.3 The pattern was striking enough at the time to attract significant attention, and yet the broader management response was modest, in part because the redesign required was substantial and in part because the underlying signals through which firms identified merit remained, as the previous essay in this series argued, calibrated to a different kind of candidate.
The ten-year review of Microsoft's own Neurodiversity Hiring Program, published by the company in April 2025 and authored by Neil Barnett, the senior director of accessibility responsible for the programme since its launch, provides a useful corrective to the optimistic version of this story. The programme, which began in 2015 with the help of Specialisterne and has since extended to a network of partners and customers, is reported to have hiring retention rates above ninety per cent for participants and consistently strong performance ratings, and the programme's frameworks have been adopted by more than five hundred organisations through Microsoft's Neurodiversity Career Connector.4 The honest reading of the same review, however, is that ten years of focused effort by one of the most capable employers in the world has produced a programme that, by Microsoft's own description, remains a programme rather than a default mode of selection. The conditions under which the rest of the firm continues to hire and evaluate have proved considerably more resistant than the dedicated programme has.
What looked like a neutral organisational template was in fact a cognitive filter, and one that admitted some kinds of mind more readily than others.
The Specific Shift that Generative Systems Introduce
The argument that distinguishes the present series from the older neurodiversity literature concerns what changes, structurally, when generative systems are placed between the cognitive work of the individual and the organisational interface that observes it. The older argument rested on the proposition that the firm should redesign itself to accommodate a wider range of minds. The newer argument is that, whether or not the firm chooses to redesign itself, the cost of producing the signals on which the existing design depends has fallen sharply, and that this collapse alters the comparative position of those who previously paid the highest price for producing those signals at all.
The most rigorous early evidence on this point comes from a 2024 study commissioned by Microsoft from EY, in which the working experience of employees with disability or neurodivergence was examined in the context of Microsoft 365 Copilot use across organisations of varying size and sector.5 The study, conducted with three hundred respondents in qualitative and quantitative phases and published as GenAI for accessibility: more human, not less, reports that ninety-two per cent of respondents found generative AI improved their working experience and that the strongest effects were reported on dimensions that matter disproportionately for neurodivergent employees, namely the cost of drafting written communication, the friction of organising multi-step tasks, the load of switching between contexts and the social effort of producing verbal summaries on demand. The companion Microsoft Research paper, Shifting Work Patterns with Generative AI, found analogous reductions in the time spent on email and document production across a broader sample, with the largest absolute time savings accruing to employees who had previously spent the most time on those activities.6
The mechanism is not that generative systems make neurodivergent employees more capable. The mechanism is that the systems reduce the production cost of the surface signals that the firm has historically used to infer capability, and that this reduction is most economically valuable for employees who previously paid the highest price to produce those signals. The point is worth stating with care, because it is too easily reduced to the more comfortable proposition that AI will help neurodivergent people fit in. The honest version is structural rather than therapeutic. The firm has been paying, throughout its history, for a particular kind of translation from internal cognition to legible output. It has been paying for that translation in two currencies, namely the time and energy of the employees doing the translating, and the foregone contribution of the employees for whom the translation cost was prohibitive. The collapse in the unit cost of that translation, which is what generative systems amount to in this domain, alters both currencies simultaneously, and the second of these alterations is the one with the larger long-term economic consequence.
The mechanism is not that generative systems make neurodivergent employees more capable. The mechanism is that the systems reduce the production cost of the surface signals that the firm has historically used to infer capability.
What the Documented Programmes Actually Show
The empirical record on dedicated neurodiversity programmes, taken seriously rather than rhetorically, supports a more measured set of conclusions than either the enthusiast or the sceptic typically draws. SAP's Autism at Work programme, launched in 2013 and reported on most extensively in the case work of Robert Austin and Thorkil Sonne, scaled to include hundreds of employees across more than a dozen countries by the second half of the 2020s, with retention rates that the company has periodically described as comparable to or higher than the broader workforce.7 The independent evaluation of DXC Technology's Dandelion Program, conducted by La Trobe University in collaboration with the Australian Government and published in successive reports between 2018 and 2024, documents productivity gains of between thirty and one hundred and forty per cent on selected technical tasks compared to neurotypical comparison groups, alongside more nuanced findings on retention, engagement and the operational cost of the support structure.8 JPMorgan Chase's Autism at Work programme, launched in 2015 and now operating in eleven countries, reports analogous findings, including, in the company's own published account, productivity comparable to or higher than that of peer cohorts on a range of technology and operations roles.9
The Auticon Global Impact Report 2025, drawing on data from four hundred and ninety-four employees, two hundred and eighty-seven client organisations and three hundred and seventy-three autistic technology consultants across fifteen countries, provides the most recent and methodologically transparent dataset on the question.10 The report, which Auticon publishes in collaboration with academic partners and which is unusually candid about the limitations of self-reported data, finds that ninety-one per cent of clients describe their autistic consultants as bringing a measurable competitive edge to their organisation, with the strongest effects reported in software quality assurance, data analytics, cybersecurity and pattern-intensive work where the cost of error is high. The same report finds that retention of autistic consultants is materially higher than the industry average for comparable roles, and that the cost of the wrap-around support model is recovered, on Auticon's accounting, through reduced rework and lower turnover within the first eighteen months of engagement.
The honest synthesis of these findings is neither that neurodivergent employees are uniformly more productive nor that they require disproportionate support. It is that, in roles where the work itself rewards sustained attention to pattern, anomaly detection and the kind of internal coherence that resists premature closure, the available evidence suggests a measurable economic surplus that has so far been captured by a small number of organisations willing to redesign the entry point and the working environment. The relevant question for the rest of the corporate sector is not whether to admire those organisations from a distance, but whether the economic case for that redesign has changed materially now that generative systems have collapsed the cost of producing the surface signals that previously kept those candidates outside the gate.
The relevant question for the rest of the corporate sector is not whether to admire those organisations from a distance, but whether the economic case for that redesign has changed materially now that generative systems have collapsed the cost of producing the surface signals that previously kept those candidates outside the gate.
A Closer Reading of the EY Centres of Excellence
The most instructive recent case is that of EY's Neuro-Diverse Centers of Excellence, which the firm has operated since 2016 and which by 2025 had grown to fifteen centres employing more than nine hundred neurodivergent professionals across the United States, the United Kingdom, India, Canada and the Philippines, with work spanning artificial intelligence development, cybersecurity, automation, blockchain assurance and audit analytics.11 The interesting feature of EY's approach, for the purposes of the present argument, is that the firm has explicitly placed the Centers at the leading edge of its own work on artificial intelligence rather than at the edge of its diversity programming. EY's collaboration with Microsoft on the development of accessible AI features in Azure, documented in a 2024 customer story published by Microsoft, was led in significant part by employees from the Neuro-Diverse Centers, on the explicit reasoning that the design of inclusive AI systems is improved when the design work is conducted by people who experience the cost of poor inclusion most acutely.12
The implication is more important than it first appears. EY has not positioned its neurodivergent workforce as a population to be supported by AI tools. It has positioned that workforce as the most credible authors of the AI tools the firm intends to build. The reordering, in this instance, runs in the opposite direction to the conventional narrative of accommodation. The firm has identified that the cognitive characteristics that previously made certain professionals difficult to integrate into a standardised office environment are precisely the characteristics that make those professionals well placed to identify, design and validate the next generation of cognitive infrastructure on which the broader workforce will increasingly depend. Whether other firms will be willing to draw the same conclusion remains an open question, but the EY case provides a useful counterexample to the assumption that neurodivergent talent belongs in a dedicated programme rather than at the centre of the firm's most consequential technical work.
EY has not positioned its neurodivergent workforce as a population to be supported by AI tools. It has positioned that workforce as the most credible authors of the AI tools the firm intends to build.
The Three Structural Shifts that Follow
Stepping back from the individual cases, the available evidence points to three structural shifts that are already underway and that will accelerate over the coming five years. Each of them changes the position of neurodivergent talent inside the firm without requiring any explicit decision by management to that effect. The first concerns the apparatus of evaluation, the second concerns the design of work itself, and the third concerns the geometry of organisational roles. Taken together, they describe a quiet reordering of talent that no diversity programme set in motion and that no diversity programme can stop.
The first shift, on evaluation, is the one developed at length in the previous essay and need not be restated here. What is worth adding, in the specific context of cognitive diversity, is that the signals on which the firm has historically relied to identify merit have been disproportionately costly for neurodivergent professionals to produce. The evaluation apparatus has therefore been imposing a tax on those professionals that did not appear on any balance sheet, and the proceeds of that tax have been collected, in the form of higher visibility and faster promotion, by professionals whose cognitive style was better aligned with the production of the signals themselves. The collapse in the cost of producing those signals does not eliminate the tax. It merely makes its existence considerably harder to obscure, since the firm can no longer point to the signals as evidence of underlying capability with the same confidence it once did.
The second shift concerns the design of work itself. The conventional unit of professional work, which has become familiar enough to feel inevitable, is the back-and-forth conversation, conducted in real time, in a room or on a video call, in which the participants are expected to formulate, articulate and refine their reasoning under conditions of social observation. The format places a premium on a particular cognitive endowment, namely the capacity to think while being watched. Generative systems, when they are integrated into the working environment with any seriousness, alter the format. They allow a larger share of the cognitive work to be done in advance, in solitude, in a written or structured form, with the synchronous meeting becoming a final calibration step rather than the primary site of thinking. This redesign, which is already visible in the working patterns of teams that have integrated Copilot, Claude or similar systems into their default workflow, restores access to a population of employees for whom the older format imposed an avoidable handicap. The firm did not redesign the work in order to include those employees. It redesigned the work for productivity reasons, and the inclusion is a structural by-product.
The third shift concerns the geometry of roles. The traditional career ladder has rewarded a single archetype, namely the broadly competent generalist whose value to the firm increases as their span of responsibility widens and whose advancement requires sustained demonstration of competence across an increasingly diverse range of social, analytical and managerial demands. The implicit assumption has been that the cognitive resources required to perform well across this range are widely distributed. The empirical record on neurodivergent professionals, as well as on the broader literature on cognitive specialisation, suggests rather that exceptional performance in a single dimension is often accompanied by ordinary or below-average performance in others, and that the conventional ladder systematically penalises this profile in favour of the more uniformly competent generalist. Generative systems, by absorbing a substantial share of the broad-spectrum cognitive load that the ladder previously required individuals to carry, make it considerably more economically rational for the firm to construct roles around specific cognitive strengths rather than around the assumption of uniform competence. The result, over time, is a flatter and more specialised organisational geometry in which the depth of a particular contribution becomes more visible and more rewardable than the breadth of the contributor.
Taken together, they describe a quiet reordering of talent that no diversity programme set in motion and that no diversity programme can stop.
Where the Promise Quietly Falters
The argument so far has been deliberately constructive, and a serious essay owes its readers a clear account of where the same forces operate against the population whose position it is supposed to be improving. The honest reading of the available evidence is that generative systems carry as much risk as opportunity for neurodivergent professionals, and that the realised outcome will depend on choices that are presently being made, often unconsciously, by the firms deploying these systems.
The most immediate risk concerns algorithmic recruitment. The Mobley v. Workday class action, in which a federal court in California granted preliminary collective certification in May 2025, alleges that Workday's AI-driven recruitment screening systematically disadvantaged applicants over the age of forty and, in subsequent filings, applicants with disabilities including neurodevelopmental conditions.13 The litigation is ongoing, and its eventual outcome will turn on questions of evidence and statutory interpretation, but the underlying mechanism is well documented in the technical literature on machine learning and is independent of the specific facts of the case. Recruitment models trained on historical hiring data inherit the selection patterns embedded in that data, and the historical record on neurodivergent candidates is one of systematic exclusion. A model trained on that record will, absent careful and ongoing intervention, reproduce the exclusion under the more impressive veneer of statistical respectability.
The HireVue case, in which the company quietly withdrew the facial-analysis component of its video interview platform in 2021 following an external audit, is instructive in a related way.14 The system had been measuring micro-expressions, vocal cadence and other markers that correlate, in the general population, with the conventions of neurotypical communication, and that diverge, in well-documented ways, in the communication style of many autistic adults. The withdrawal of the facial-analysis component was a useful corrective, but the underlying logic of inferring capability from the surface texture of communication is preserved across the sector. Generative systems, by lowering the cost of producing that surface texture for those who can adapt their output to the format the system rewards, raise the relative cost of being a candidate whose surface texture does not adapt readily, regardless of underlying capability.
The third risk is more subtle and concerns the gradual narrowing of cognitive variation inside the firm. The Doshi and Hauser study, published in Science Advances in July 2024, found that generative systems enhance individual creativity while reducing the collective diversity of what is produced, since outputs converge towards the implicit aesthetic of the underlying model.15 The implication, applied to a workforce that integrates generative systems pervasively, is that the variation in output that the firm observes is increasingly variation in prompting strategy rather than variation in underlying cognition. The neurodivergent employee whose distinctive contribution lies precisely in the unconventional structure of their thinking risks finding that the unconventional structure is smoothed away by the same systems that make their work more legible, with the firm receiving a more polished but less distinctive contribution and rewarding the polish at the expense of the distinctiveness. The risk is real, and it requires the firm to take an active interest in preserving the cognitive variation that the systems are not designed to preserve on its behalf.
A model trained on that record will, absent careful and ongoing intervention, reproduce the exclusion under the more impressive veneer of statistical respectability.
A Note on the Daniela Amodei Counter-Position
The argument of this essay, and of the series more broadly, has occasionally been read as a prediction that neurodivergent professionals will be disproportionate beneficiaries of the next phase of organisational life. That is not the argument, and the honest version requires engagement with the most credible counter-position, which has been articulated most clearly by Daniela Amodei, the president of Anthropic, in a series of interviews and conference appearances during 2025. Amodei has argued that as generative systems absorb a larger share of the analytical and pattern-recognition work that has historically been the comparative strength of certain neurodivergent profiles, the residual differentiator inside the firm shifts towards capabilities that the same systems do not perform well, namely emotional attunement, social judgement and the management of ambiguity in human interaction.16
The counter-position deserves serious consideration rather than dismissal. The empirical literature on the comparative emotional load of contemporary office life on autistic and ADHD professionals is consistent enough that any honest projection has to take it into account. The argument of the present essay is not that the structural shift will be uniformly favourable to neurodivergent talent. The argument is that the realised distribution of outcomes will depend on which dimensions of work the firm decides to evaluate and reward, and that the firm's current evaluation apparatus is poorly equipped to distinguish between the dimensions whose value is rising and those whose value is falling. A firm that responds to the Amodei argument by doubling down on its existing measurement of social fluency, without simultaneously revising its measurement of analytical depth and pattern recognition under the new conditions, will produce an outcome that vindicates neither the optimistic nor the pessimistic reading of the shift, but that accelerates the loss of the cognitive variation it can least afford to lose.
A firm that responds to the Amodei argument by doubling down on its existing measurement of social fluency, without simultaneously revising its measurement of analytical depth and pattern recognition under the new conditions, will accelerate the loss of the cognitive variation it can least afford to lose.
A Word on the Karp Survivorship Argument
At the other end of the public debate sits the position articulated by Alex Karp of Palantir, who has argued in his recent writing and public appearances that the gifted neurodivergent profile, which he sometimes describes in the language of cognitive elitism, will be the dominant beneficiary of the new conditions and that the broader workforce should expect a steeper hierarchy in which a small number of exceptional minds capture a disproportionate share of organisational value.17 The Karp position has the rhetorical advantage of being clear and the analytical disadvantage of being a survivorship argument. It generalises from the experience of the small number of neurodivergent professionals who have already cleared every filter the modern firm imposes, and it draws conclusions about the population from observations on the survivors. The available data on the broader neurodivergent workforce, including the EY 2025 study and the Auticon 2025 impact report cited above, point to a considerably more heterogeneous population whose realised outcomes depend less on innate cognitive endowment than on whether the firm has been willing to redesign the conditions under which that endowment becomes economically visible.
The honest position, which lies between the two public extremes, is that the structural shift creates the possibility of a quiet reordering in which neurodivergent talent moves from the margins of the firm towards the centre of its most consequential technical work, and that whether the possibility is realised will depend on a series of unglamorous decisions about evaluation, work design and role geometry that very few firms are presently making with any seriousness. The present essay closes the series with an attempt to articulate, in concrete terms and for an executive audience, what those decisions look like.
The Karp position has the rhetorical advantage of being clear and the analytical disadvantage of being a survivorship argument. It generalises from the experience of the small number of neurodivergent professionals who have already cleared every filter the modern firm imposes, and it draws conclusions about the population from observations on the survivors.
The Strategic Question for Executives
The strategic question that the foregoing analysis presents to a chief executive is not whether the firm should expand its diversity programming. The question is whether the firm has correctly identified the population on whose cognitive contribution its competitive position over the next decade will increasingly depend, and whether the apparatus by which it currently selects, evaluates and promotes its workforce is calibrated to recognise that contribution in the conditions that are now emerging rather than in the conditions that used to obtain. The honest answer, in most firms, is that the apparatus has not been examined in this light, and that the cost of the examination has been postponed in favour of the more familiar work of integrating generative systems into existing workflows without disturbing the organisational architecture that the workflows assumed.
The first concrete decision concerns the firm's recruitment process. A firm that wishes to retain access to the cognitive variation that will increasingly be its differentiator will need to examine the early-stage filters of its hiring funnel, including any algorithmic screening, structured interview formats and group assessment exercises, and ask the unromantic question of what those filters were originally designed to measure and whether the measurement remains valid under conditions in which the production cost of the surface signals has collapsed. The likely conclusion, in most firms, is that several of the filters were measuring the cost of legibility rather than the underlying capability, and that the firm has been systematically excluding candidates whose internal cognitive endowment was strong but whose surface presentation was costly to produce. The remedy is not a specialised neurodiversity hiring programme bolted onto the existing process. The remedy is a redesign of the underlying process such that the dedicated programme becomes redundant.
The second concrete decision concerns the format of evaluation. A firm that continues to evaluate its employees primarily through synchronous behavioural observation, calibrated against the conventions of neurotypical communication, will find that its evaluation outputs become progressively less informative as generative systems absorb the production cost of the very communication being observed. The remedy is to shift a greater share of the evaluation weight onto outcome measures that the systems do not produce on the employee's behalf, namely the quality of the underlying judgement, the appropriateness of the problem framing and the durability of the contribution over a longer time horizon. This shift, which has been recommended by the structured-evaluation literature for decades, has been resisted on grounds of administrative cost. The collapse of the legibility tax removes the most credible defence of the resistance.
The third concrete decision concerns the geometry of roles. A firm that constructs its career ladder around the implicit assumption of uniform competence across analytical, social and managerial dimensions will find that the assumption becomes progressively more expensive to defend as generative systems absorb a larger share of the broad-spectrum load. The remedy is to construct, alongside the conventional ladder, a parallel track for deep specialists whose contribution is compensated and recognised on terms appropriate to its character, rather than measured against the uniform-competence template. The pattern is familiar from the engineering ladders of large technology firms and from the partnership structures of professional services. What the present argument adds is that the parallel track becomes economically more important once the broad-spectrum load has been partially externalised to synthetic systems, since the residual value of the human contribution shifts disproportionately towards depth.
The fourth concrete decision concerns the working environment. A firm that has invested in open-plan offices, ambient social pressure and the implicit expectation of constant social availability has been imposing, throughout that period, a cognitive load that falls disproportionately on neurodivergent employees and that has been documented in the literature as a measurable productivity drag on the broader workforce as well.18 The integration of generative systems into the working day does not, by itself, alter the environment, but it does alter the economic case for retaining an environment that imposes a translation cost the systems are now in a position to remove. A firm that wishes to benefit fully from the structural shift will need to revisit the physical and temporal architecture of its working environment with the same seriousness with which it currently revisits its software stack.
The fifth and most uncomfortable decision concerns the firm's account of merit. As long as the firm continues to describe its promotion outcomes in the vocabulary of capability, while in fact selecting on the surface signals whose informational content has eroded, the gap between the public account and the internal mechanism will widen. The widening gap carries reputational, legal and operational risk, of which the legal component has begun to manifest in the recent class-action litigation around algorithmic hiring and is unlikely to remain confined to that domain. A firm that wishes to preserve the integrity of its account will need to revise the account, not by abandoning the language of merit but by being considerably more precise about the dimensions of capability the firm is in fact prepared to measure and reward in the new conditions. The work is not technical, it is institutional, and it cannot be delegated.
The remedy is not a specialised neurodiversity hiring programme bolted onto the existing process. The remedy is a redesign of the underlying process such that the dedicated programme becomes redundant.
The Closing Frame
The series began with the observation that for decades organisations optimised processes, workflows and incentives without questioning the cognitive assumptions embedded within them. The intervening essays traced the consequences of those assumptions, from the economic visibility of cognitive styles previously suppressed by the cost of legibility, through the redistribution of executive function, the rise of synthetic professionalism, the honest economics of neurodivergent contribution, the linearity assumption embedded in organisational design, the misfiring of recruitment under the new cost structure and, in the immediately preceding essay, the illusion of meritocratic selection that has continued to operate in conditions for which it was not designed. The capstone, attempted here, has been to draw the threads together around the population whose position is most directly altered by the structural shift the series has been describing throughout.
The reordering of talent that this essay has tried to describe will not announce itself. It will not appear in the firm's diversity reports, since it is not a programme. It will not appear in the firm's productivity statistics, since the productivity gains will be attributed to the systems rather than to the cognitive variation the systems have made newly visible. It will appear, instead, in the slow drift of which firms attract and retain the cognitive endowments on which the most consequential work increasingly depends, and in the slow drift of which firms continue to mistake the appearance of capability, now produced more cheaply than ever, for the underlying contribution they are in fact paying for. The drift is already underway, and the firms that recognise it earliest will have the largest opportunity to act on it before it becomes a competitive necessity rather than a strategic choice.
The series closes here, in the sense that the argument has reached the population whose position depends most directly on it. The argument itself, however, is at the beginning rather than the end of its working life, and the invitation that the present essay extends to its readers is not to admire the analysis but to test it, with reference to their own firm, against the evidence available to them and the decisions presently in front of them. The reordering will continue regardless of who chooses to act on it. The question, for the executives reading these pages, is whether their firm will recognise the shift in time to participate in it on terms favourable to the institution, or whether it will discover the shift retrospectively, in the form of contributions it can no longer attract and contributions it is no longer able to recognise inside its own walls.
The reordering will continue regardless of who chooses to act on it. The question, for the executives reading these pages, is whether their firm will recognise the shift in time to participate in it on terms favourable to the institution, or whether it will discover the shift retrospectively, in the form of contributions it can no longer attract and contributions it is no longer able to recognise inside its own walls.
Questions this essay answers
What is the central argument of the capstone essay?
The central argument is that the unit cost of producing the signals through which firms infer competence has fallen sharply, that this fall reorders the comparative position of cognitive styles whose translation cost was previously prohibitive and that the resulting redistribution is structural rather than therapeutic and is already under way.
Which empirical sources are most load-bearing?
The most load-bearing sources are Austin and Pisano in Harvard Business Review in 2017, the EY Global Neuroinclusion at Work Study published in July 2025, the Microsoft and EY Copilot study published in 2024, the Microsoft Research paper on shifting work patterns and the published programme reviews from SAP, Microsoft, JPMorgan Chase, EY Neuro-Diverse Centers of Excellence, DXC Dandelion as evaluated by La Trobe University and the Auticon Global Impact Report 2025.
What should executives do in the next twelve months?
Executives should commission an honest review of how their selection, evaluation and work-design processes interact with the new cost structure of signal production, and should treat the redesign of those three areas as a strategic question about access to talent rather than as an adjacent inclusion initiative.
Sources
- Deloitte Insights, "A rising tide lifts all boats: Creating a better work environment for all by embracing neurodiversity," Deloitte Review, Issue 30, January 2022; SAP, "Autism at Work: Frequently Asked Questions," SAP corporate communications, updated 2024; World Economic Forum, "Why neurodiversity is the next frontier in workplace inclusion," WEF Insight Report, 2023.
- EY Global, Global Neuroinclusion at Work Study 2025: How can action on neuroinclusion accelerate business transformation?, July 2025, available at ey.com.
- Austin, R. D. and Pisano, G. P., "Neurodiversity as a Competitive Advantage," Harvard Business Review, May–June 2017, pp. 96–103.
- Barnett, N., "A Decade of Learning: Building a Dynamic Workforce through Neurodiversity," Microsoft Accessibility Blog, April 2025.
- EY and Microsoft, GenAI for accessibility: more human, not less. How does Microsoft 365 Copilot impact the working experience of people living with disability and/or neurodivergence?, EY Global, December 2024.
- Microsoft Research, "Shifting Work Patterns with Generative AI," Microsoft Research Publication, 2024; companion article on the AI Data Drop platform, Microsoft WorkLab, December 2024.
- Austin, R. D. and Sonne, T., "The Dandelion Principle: Redesigning Work for the Innovation Economy," MIT Sloan Management Review, Vol. 55, No. 4, Summer 2014; SAP, "Autism at Work programme update," SAP corporate communications, 2024; Bluewave Select, The Neurodiversity in SAP Report 2024, January 2024.
- Hedley, D., Wilmot, M., Spoor, J. and Dissanayake, C. et al., "Evaluation of the DXC Dandelion Employment Program," Olga Tennison Autism Research Centre, La Trobe University, in collaboration with the Australian Government Departments of Human Services, Defence, and Immigration and Border Protection, 2018, with subsequent updates 2020 and 2024; DXC Technology, Social Impact Practice Year in Review 2024, DXC corporate publication.
- JPMorgan Chase, "Reaching out to the brightest new technologists across the neurodivergent spectrum," JPMorgan Chase Newsroom, with subsequent annual updates through 2025; Annabi, H., Crooks, E., Barnett, N., Guadagno, J., Mahoney, J. R. and Pacilio, A., Autism @ Work Playbook, ACCESS-IT, University of Washington Information School, 2019, drawing on the experience of SAP, Microsoft, JPMorgan Chase, EY and others.
- Auticon, The Edge Effect: auticon's Global Impact Report 2025, Auticon GmbH, 2025; earlier Auticon Global Impact Report 2022 for trend comparison.
- EY Global, "EY Neuro-Diverse Centers of Excellence," EY corporate communications, updated 2025; Austin, R. D., "Tapping the Talent of People with Autism: SAP, JPMorgan Chase and EY," Ivey Business School Case 9B19C019, Ivey Publishing, 2019, with subsequent updates.
- Microsoft Customer Stories, "EY helps enhance the inclusivity of AI development with Azure AI," Microsoft, 19 November 2024.
- Mobley v. Workday, Inc., Case No. 23-cv-00770-RFL, United States District Court, Northern District of California; order on motion to dismiss, July 2024; order granting preliminary collective certification, May 2025; subsequent procedural filings 2025–2026.
- Kahn, J., "HireVue drops facial monitoring amid A.I. algorithm audit," Fortune, 19 January 2021; Engler, A., "Auditing employment algorithms for discrimination," Brookings Institution, March 2021.
- Doshi, A. R. and Hauser, O. P., "Generative AI enhances individual creativity but reduces the collective diversity of novel content," Science Advances, Vol. 10, eadn5290, July 2024.
- Amodei, D., interviews and public remarks during 2025 on the comparative durability of human emotional and social judgement under generative AI, including coverage in The Wall Street Journal and Bloomberg; complementary discussion in Anthropic, The Economic Index, periodic releases 2024–2025.
- Karp, A. and Zamiska, N., The Technological Republic: Hard Power, Soft Belief, and the Future of the West, Crown Currency, 2025; subsequent interviews and conference appearances by Karp throughout 2025.
- Bernstein, E. S. and Turban, S., "The impact of the 'open' workspace on human collaboration," Philosophical Transactions of the Royal Society B, Vol. 373, 20170239, July 2018; complementary findings in subsequent literature on cognitive load in open-plan environments.
About this Series
This article concludes the Synthetic Minds at Work series. The structural questions it raises continue in the Decision Architecture series, in particular in the analysis of how organisations design the conditions under which judgement is exercised and authority is allocated.
