The Paradox of Professional Input: How Expert Collaboration with AI Systems Shapes Their Future Value

Krishna Kumar Balaraman; Venkat Ram Reddy Ganuthula

arxiv: 2504.12654 · v1 · submitted 2025-04-17 · 💰 econ.GN · cs.AI· q-fin.EC

The Paradox of Professional Input: How Expert Collaboration with AI Systems Shapes Their Future Value

Venkat Ram Reddy Ganuthula , Krishna Kumar Balaraman This is my paper

Pith reviewed 2026-05-22 20:12 UTC · model grok-4.3

classification 💰 econ.GN cs.AIq-fin.EC

keywords human-AI collaborationtacit knowledge externalizationprofessional expertiseautomation paradoxknowledge managementlabor economicsAI systemsexpertise evolution

0 comments

The pith

As experts externalize tacit knowledge to AI, they may automate their own roles while creating opportunities for new forms of professional value.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This perspective paper examines the paradox that arises when professionals collaborate with AI systems by making their implicit expertise explicit. The analysis across multiple contexts suggests that this externalization can accelerate automation of traditional tasks. At the same time, it opens avenues for expertise to evolve into new, higher-value activities that humans are better positioned to perform. The authors draw on knowledge management and labor economics to argue that thoughtful navigation of this process can preserve and transform professional worth rather than diminish it. They outline steps for education, organizations, and policy to steer the outcome toward enhancement of human roles.

Core claim

Through collaboration with AI, domain experts externalize their tacit knowledge, which risks automating their current expertise yet simultaneously generates possibilities for the evolution of that expertise into novel forms of professional value that remain distinct from what machines can replicate.

What carries the argument

The externalization of tacit knowledge in human-AI collaboration, which acts as both a driver of potential automation and a catalyst for redefining professional contributions.

If this is right

Professionals will shift emphasis from performing routine expert tasks to overseeing, refining, and innovating around AI outputs.
Organizations will redesign roles and workflows to integrate AI codification while keeping space for human judgment in ambiguous cases.
Professional education programs will incorporate training in knowledge externalization and AI partnership skills.
Policy will focus on incentives that reward the creation of new human-only value layers rather than pure efficiency gains.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Fields with strong existing communities of practice may adapt faster by collectively deciding what knowledge to externalize and what to retain as human advantage.
The paradox could widen gaps between adaptable experts who learn to direct AI and those who do not, affecting wage distributions within professions.
Over time this dynamic might favor hybrid human-AI systems where the human contribution is precisely the part that cannot be fully codified.

Load-bearing premise

That patterns of collaboration seen in studied professions will hold broadly and that sharing implicit knowledge will mainly speed automation without major mechanisms that protect or increase human professional value.

What would settle it

A study tracking earnings, employment rates, and task composition over ten years in fields with high versus low rates of expert knowledge codification for AI training.

read the original abstract

This perspective paper examines a fundamental paradox in the relationship between professional expertise and artificial intelligence: as domain experts increasingly collaborate with AI systems by externalizing their implicit knowledge, they potentially accelerate the automation of their own expertise. Through analysis of multiple professional contexts, we identify emerging patterns in human-AI collaboration and propose frameworks for professionals to navigate this evolving landscape. Drawing on research in knowledge management, expertise studies, human-computer interaction, and labor economics, we develop a nuanced understanding of how professional value may be preserved and transformed in an era of increasingly capable AI systems. Our analysis suggests that while the externalization of tacit knowledge presents certain risks to traditional professional roles, it also creates opportunities for the evolution of expertise and the emergence of new forms of professional value. We conclude with implications for professional education, organizational design, and policy development that can help ensure the codification of expert knowledge enhances rather than diminishes the value of human expertise.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This is a balanced literature synthesis on AI-expert collaboration that flags both risks and opportunities but adds no new data, models, or tests.

read the letter

The core takeaway is that this perspective paper organizes existing ideas from knowledge management, expertise studies, HCI, and labor economics around the idea that externalizing tacit knowledge to AI can both erode and reshape professional value. It proposes navigational frameworks without claiming to resolve the tension empirically. That framing is clear and even-handed in the abstract. The paper does a decent job pulling patterns across professional contexts and avoiding one-sided alarmism about automation. The conclusion on implications for education, organizations, and policy follows logically from the synthesis it cites. Those are real strengths for a non-empirical piece. The main limitation is that everything rests on interpretive reading of prior work rather than fresh evidence or derivations. No original datasets, no parameter-free predictions, and no formal checks on whether the observed patterns generalize or whether offsetting mechanisms dominate. The weakest assumption—that collaboration patterns are broadly representative—remains untested here, which is fine for a perspective but caps how far the claims can travel. The citation pattern looks standard for this kind of overview and does not appear circular. This is useful for readers already working on AI labor issues who want a compact map of the arguments, but it will not move the research frontier. A journal that accepts perspectives could reasonably send it for review to see whether the frameworks add enough structure to justify publication; a data-driven or theory journal should desk-reject.

Referee Report

1 major / 2 minor

Summary. This perspective paper examines the paradox whereby domain experts collaborating with AI systems by externalizing tacit knowledge may accelerate the automation of their own expertise, while simultaneously creating opportunities for evolved forms of professional value. Drawing on syntheses of literature in knowledge management, expertise studies, HCI, and labor economics, the manuscript identifies emerging patterns in human-AI collaboration across professional contexts and proposes navigational frameworks, concluding with implications for professional education, organizational design, and policy.

Significance. If the interpretive synthesis holds, the paper provides a balanced lens on AI's dual effects on professional expertise, emphasizing preservation and transformation of human value rather than outright displacement. Its contribution lies in integrating cross-disciplinary insights to outline actionable frameworks, which could inform adaptation strategies for professionals and institutions amid advancing AI capabilities.

major comments (1)

[Analysis of multiple professional contexts] The central claim that observed patterns of human-AI collaboration are broadly representative across professional contexts (as synthesized in the analysis of multiple contexts) rests on interpretive generalization without original empirical validation or falsifiable tests; this assumption underpins the proposed frameworks but lacks discussion of potential counterexamples or sector-specific variations that could limit generalizability.

minor comments (2)

[Proposed frameworks] The navigational frameworks would benefit from more explicit mapping to the cited literature streams to clarify how each element derives from prior work in knowledge management or labor economics.
[Emerging patterns] Consider adding a brief table or structured summary of the identified patterns to improve readability and allow readers to trace the synthesis more directly.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive comments on our perspective paper. We address the single major comment below and outline revisions that will strengthen the manuscript's transparency regarding its interpretive nature and scope.

read point-by-point responses

Referee: [Analysis of multiple professional contexts] The central claim that observed patterns of human-AI collaboration are broadly representative across professional contexts (as synthesized in the analysis of multiple contexts) rests on interpretive generalization without original empirical validation or falsifiable tests; this assumption underpins the proposed frameworks but lacks discussion of potential counterexamples or sector-specific variations that could limit generalizability.

Authors: We agree that the manuscript is an interpretive synthesis rather than an empirical study and therefore does not offer original data, falsifiable tests, or statistical validation of generalizability. The patterns discussed are drawn from secondary sources across knowledge management, expertise studies, HCI, and labor economics. To address the concern, we will add an explicit subsection on scope and limitations that (a) states the interpretive character of the synthesis, (b) notes the absence of sector-specific empirical tests, and (c) identifies plausible counterexamples and contextual variations (e.g., highly regulated fields such as medicine versus creative industries) that could moderate the observed patterns. This addition will clarify the boundaries of the proposed frameworks without altering the paper's perspective format. revision: yes

Circularity Check

0 steps flagged

No significant circularity in perspective synthesis

full rationale

The manuscript is explicitly a perspective paper that synthesizes existing literature from knowledge management, expertise studies, HCI, and labor economics to identify patterns and propose navigational frameworks. It advances no formal derivations, equations, empirical models, or falsifiable predictions that could reduce by construction to fitted parameters or self-citations. The central claim is balanced and interpretive, acknowledging both risks and opportunities without load-bearing premises that collapse into the paper's own inputs. As a result the analysis remains self-contained against external benchmarks and receives a score of 0.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper is a perspective analysis relying on background assumptions from knowledge management and expertise literature rather than introducing quantitative models, free parameters, or new entities.

axioms (2)

domain assumption Domain experts hold significant implicit or tacit knowledge that can be externalized when collaborating with AI systems.
This premise underpins the identified paradox and is invoked in the abstract's description of collaboration dynamics.
ad hoc to paper Externalization of tacit knowledge through AI collaboration accelerates automation of professional expertise.
This is presented as the core mechanism creating risks to traditional roles.

pith-pipeline@v0.9.0 · 5696 in / 1426 out tokens · 103696 ms · 2026-05-22T20:12:06.653722+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages · 2 internal anchors

[1]

expertise externalization paradox,

Title: The Paradox of Professional Input: How Expert Collaboration with AI Systems Shapes Their Future Value Venkat Ram Reddy Ganuthula1 Krishna Kumar Balaraman1 1Indian Institute of Technology Jodhpur Abstract This perspective paper examines a fundamental paradox in the relationship between professional expertise and artificial intelligence: as domain ex...

work page 2019
[2]

and Med-PaLM 2 (Singhal et al., 2023), exhibit remarkable capabilities in contextual reasoning and domain-specific tasks. These developments drive adoption across healthcare (Esteva et al., 2019; Topol, 2019), legal services (Alarie, Niblett, & Yoon, 2018; McGinnis & Pearce, 2014), financial analysis (Cao, 2022; López de Prado, 2018), and creative industr...

work page 2023
[3]

As professionals refine these tools, they externalize relational implicit knowledge (Collins, 2010), reshaping the organization of knowledge work

illustrate AI’s growing reach while McKinsey (2025a) estimates a $4.4 trillion productivity boost from AI use cases, underscoring its economic impact. As professionals refine these tools, they externalize relational implicit knowledge (Collins, 2010), reshaping the organization of knowledge work. Drawing on Christensen’s (1997) disruptive innovation theor...

work page 2010
[4]

Understanding this dynamic is vital for professionals, organizations, and policymakers navigating AI’s integration into knowledge-intensive fields

forecasts 170 million new jobs and 92 million displaced by 2030, netting 78 million jobs, further illustrating AI’s dual impact. Understanding this dynamic is vital for professionals, organizations, and policymakers navigating AI’s integration into knowledge-intensive fields. In the following sections, we define professional expertise, distinguishing expl...

work page 2030
[5]

we can know more than we can tell,

captured this with his observation, "we can know more than we can tell," highlighting intuitions, contextual awareness, and pattern recognition honed through experience. Collins (2010) refines this into "relational tacit knowledge" (transferable via social interaction, e.g., a clinician’s diagnostic hunch based on patient cues) and "somatic tacit knowledg...

work page 2010
[6]

gut feel

or architects adjust designs mid-process (Cross, 2004). Studies confirm this across fields: physicians blend analytical and intuitive reasoning (Norman et al., 2007), while executives rely on "gut feel" for strategy (Sadler-Smith & Shefy, 2004). Traditionally, tacit knowledge develops through socially embedded mechanisms. Lave and Wenger’s (1991) "legitim...

work page 2004
[7]

explicitation

or Med-PaLM 2 (Singhal et al., 2023), embedding diagnostic pattern recognition. Lawyers highlight relevant passages in legal documents to enhance search algorithms, as with GPT-4 applications (Grabmair et al., 2015; Zhong et al., 2020; OpenAI, 2023), while designers select preferred outputs from generative models like Adobe Sensei to refine aesthetic para...

work page 2023
[8]

In medicine, clinical expertise externalizes through multiple AI interactions

over somatic forms. In medicine, clinical expertise externalizes through multiple AI interactions. Diagnostic systems for retinal disease (De Fauw et al., 2018), dermatology (Esteva et al., 2017), and radiology (McKinney et al.,

work page 2018
[9]

leverage such inputs. Clinicians’ corrections—e.g., overriding AI suggestions in diabetic retinopathy screening (Beede et al., 2020)—refine algorithms for edge cases, while integration with electronic health records (EHRs) enables continuous learning from decisions (Rajkomar et al., 2018; Chen et al., 2020). A 2024 study found 75% of U.S. hospitals using ...

work page 2020
[10]

clinical intuition

with Przegalinska et al. (2025) noting AI enhances clinical outcomes through task-technology fit, embedding relational tacit knowledge into scalable systems. These interactions embed relational tacit knowledge into scalable systems, though somatic skills like physical examination resist codification. This shift challenges medicine’s reliance on tacit expe...

work page 2025
[11]

or refining GPT-4-generated contracts (Alarie et al., 2018; OpenAI,

work page 2018
[12]

encode judgment about relevance and phrasing. Routine research tasks—e.g., prioritizing sources (Grabmair et al., 2015; Zhong et al., 2020)—feed search algorithms, with a 2024 ABA survey reporting 80% of firms using AI weekly for such purposes (ABA, 2024). These actions transfer relational tacit knowledge, enabling systems to mimic case-specific reasoning...

work page 2015
[13]

Junior attorneys traditionally learned from seniors, funded by billable hours (Henderson, 2013; Ribstein, 2010)

faces disruption. Junior attorneys traditionally learned from seniors, funded by billable hours (Henderson, 2013; Ribstein, 2010). AI’s codification of senior judgment—e.g., contract drafting—empowers less experienced practitioners (McGinnis & Pearce, 2014; Susskind, 2013), compressing hierarchies. BLS (2025a) data show a 12% drop in paralegal jobs from 2...

work page 2013
[14]

Writers editing GPT-4 text (Radford et al., 2019; OpenAI,

embed preferences into algorithms. Writers editing GPT-4 text (Radford et al., 2019; OpenAI,

work page 2019
[15]

Continuous feedback loops—e.g., 70% of designers tweaking AI outputs daily (AIGA, 2024)—codify relational tacit knowledge, enabling autonomous generation

teach style and tone, with Sudowrite supporting 15,000+ authors by 2024 (Clark et al., 2018; Sudowrite, 2024). Continuous feedback loops—e.g., 70% of designers tweaking AI outputs daily (AIGA, 2024)—codify relational tacit knowledge, enabling autonomous generation. McKinsey (2025a) highlights AI’s $4.4 trillion productivity potential, partly from creative...

work page 2024
[16]

or wealth managers tailoring robo-advisor plans (D’Acunto et al., 2019; Lam,

work page 2019
[17]

market sense

encode market judgment. Integration into workflows—e.g., 85% of U.S. trading firms using AI by 2024 (FINRA, 2024)—facilitates learning from decisions (Agrawal et al., 2019; Hammond, 2016). BLS (2025b) projects a 9.5% growth for financial analysts from 2023–2033, indicating augmentation over displacement, while CEPR (2025) reports a 0.5-0.6% productivity b...

work page 2024
[18]

de-skilling

shifts attorney focus to negotiation and advocacy, per a 2024 ABA survey noting 60% of firms reallocating hours (ABA, 2024; Susskind, 2013). Task restructuring carries dual implications for expertise. It may elevate the value of tacit knowledge resistant to automation—like interpersonal skills and ethical judgment (Deming, 2017; Frey & Osborne, 2017)—with...

work page 2024
[19]

social skills premium

with McKinsey (2025a) noting $4.4 trillion in productivity gains amplifying top performers. CEPR (2025) confirms a 0.5-0.6% productivity boost in Japan, suggesting broader economic shifts (CEPR, 2025). A 2024 Bloomberg report notes 10% of financial analysts shifting to oversight roles since 2020 (Bloomberg, 2024), amplifying this trend. Professional value...

work page 2025
[20]

automation paradox

but may shrink future demand (Autor, 2015; Susskind & Susskind, 2015). McKinsey (2025b) notes only 1% of firms are AI-mature, suggesting early-stage productivity gains still favor collaboration. Zuboff’s (1988) "automation paradox" pits short-term gains against long-term security, varying by career stage and specialization. A 2024 survey found 60% of mid-...

work page 2015
[21]

stepping up

while leveraging somatic and contextual strengths. These strategies balance collaboration with AI against potential replacement, offering practical pathways for sustained relevance. The first response, termed "stepping up" by Davenport and Kirby (2016), shifts professionals to higher abstraction and oversight as routine tasks automate. While AI excels in ...

work page 2016
[22]

explainable AI literacy

add significant value, with 65% of U.S. hospitals mandating such oversight by 2024 (HIMSS, 2024). Bailey and Barley (2020) note data scientists bridging technical and domain roles thrive, a model applicable across fields. Developing supervisory expertise demands educational shifts beyond traditional knowledge. Hoffman et al.’s (2017) "explainable AI liter...

work page 2024
[23]

deliberative rationality

to sustain tacit knowledge resisting externalization. Historically, these groups nurtured expertise through social participation (Lave & Wenger, 1991), and amid AI’s rise, they can prioritize human strengths (Bailey & Barley, 2020; Zuboff, 1988). This includes Dreyfus and Dreyfus’ (1986) "deliberative rationality"—critical reflection for novel contexts—an...

work page 1991
[24]

design thinking

that blend judgment with augmentation. Value shifts from information control to contextual application and trust-building (Abbott, 1988; Susskind & Susskind, 2015). For instance, 80% of lawyers in a 2024 ABA survey saw AI as enhancing, not replacing, client counsel (ABA, 2024). Schön’s (1983) "design thinking" aids problem-framing, while ethics evolve—e.g...

work page 1988
[25]

teaming structures

shifts to systems while somatic and contextual expertise persists. At the organizational level, externalization necessitates rethinking knowledge management, work design, and professional development. Traditional structures assume distinct expert-novice boundaries, with expertise accruing along defined career paths (Abbott, 1988; Freidson, 1970). As AI co...

work page 1988
[26]

shadow learning

counter this, fostering tacit acquisition; e.g., 65% of architects in 2024 valued peer forums for design intuition (AIA, 2024). Beane’s (2019) "shadow learning"—like Johns Hopkins’ surgical labs (Topol, 2019)—and Edmondson’s (1999) "psychological safety" ensure professionals critique AI limits, with 60% of engineers reporting safer innovation spaces by 20...

work page 2024
[27]

expert systems thinking

reflects this, valuing navigation of complexity over mere processing (Bailey & Barley, 2020; Pasquale, 2020). Feasibility hinges on scalable programs, though resource disparities challenge smaller firms (Brynjolfsson et al., 2018). At the educational level, the paradox upends traditional training and credentialing. Balancing theory and practice (Eraut, 20...

work page 2020
[28]

technological gatekeeping

exemplify this, integrating human-tech judgment (Davenport & Kirby, 2016). At the policy level, the paradox raises questions of regulation, labor protections, and expertise access. Regulation historically ensures quality and autonomy (Abbott, 1988; Freidson, 1970), now requiring "technological gatekeeping" (Susskind & Susskind, 2015)—defining AI autonomy ...

work page 2016
[29]

Human-in-the-loop

guide high-stakes use; the EU’s 2024 AI Act mandates human review in healthcare (EU, 2024). "Human-in-the-loop" rules (Pasquale,

work page 2024
[30]

New social contracts

persist, with 75% of U.S. hospitals enforcing this by 2024 (HIMSS, 2024). Labor policies must adapt to disrupted career paths. "New social contracts" (Brynjolfsson & McAfee,

work page 2024
[31]

countervailing investments

share tech gains equitably, with "countervailing investments" (Acemoglu & Restrepo, 2019)—e.g., $600 million in U.S. AI training grants by 2024 (DOL, 2024)—creating opportunities. "Institutional innovations" (Autor,

work page 2019
[32]

Appropriate use

like tax incentives aid transitions, though funding lags (BLS, 2024). Access policies balance AI’s democratizing potential—25% legal cost cuts since 2022 (ABA, 2024)—with quality concerns. "Appropriate use" guidelines (Susskind,

work page 2024
[33]

These responses acknowledge externalization as a social shift, not just a technical one

ensure accountability, as in California’s 2024 AI liability laws (CA, 2024). These responses acknowledge externalization as a social shift, not just a technical one. Preserving human judgment, creativity, and ethics—via frameworks balancing AI and human strengths— aims for a future where professional expertise is enhanced, not diminished, despite implemen...

work page 2024
[34]

On the Opportunities and Risks of Foundation Models

or lawyers refining GPT-4 (OpenAI, 2023)—in a fluid, often unintended process (Amershi et al., 2014). A 2024 BLS survey notes 70% of U.S. professionals use AI daily, amplifying this shift (BLS, 2024), with implications often unrecognized until deployment. Second, it varies by domain: medicine’s diagnostic shift with 75% of hospitals using AI-enhanced EHRs...

work page internal anchor Pith review Pith/arXiv arXiv 2023
[35]

S., Shulman, L

Elstein, A. S., Shulman, L. S., & Sprafka, S. A. (1978). Medical problem solving: An analysis of clinical reasoning. Harvard University Press. Eraut, M. (2000). Non-formal learning and tacit knowledge in professional work. British Journal of Educational Psychology, 70(1), 113–136. Esteva, A., Kuprel, B., Novoa, R. A., Ko, J., Swetter, S. M., Blau, H. M., ...

work page arXiv 1978
[36]

CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning

Rajkomar, A., Oren, E., Chen, K., Dai, A. M., Hajaj, N., Hardt, M., … & Dean, J. (2018). Scalable and accurate deep learning with electronic health records. npj Digital Medicine, 1(1), 1–10. Rajpurkar, P., Irvin, J., Zhu, K., Yang, B., Mehta, H., Duan, T., … & Ng, A. Y. (2017). CheXNet: Radiologist-level pneumonia detection on chest X-rays with deep learn...

work page internal anchor Pith review Pith/arXiv arXiv 2018
[37]

[https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work] McKinsey (2025b)

McKinsey & Company. [https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work] McKinsey (2025b). The state of AI: How organizations are rewiring to capture value. McKinsey & Company. [https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-a...

work page 2025
[38]

[https://www.weforum.org/publications/future-of-jobs-report-2025/]

World Economic Forum. [https://www.weforum.org/publications/future-of-jobs-report-2025/]

work page 2025

[1] [1]

expertise externalization paradox,

Title: The Paradox of Professional Input: How Expert Collaboration with AI Systems Shapes Their Future Value Venkat Ram Reddy Ganuthula1 Krishna Kumar Balaraman1 1Indian Institute of Technology Jodhpur Abstract This perspective paper examines a fundamental paradox in the relationship between professional expertise and artificial intelligence: as domain ex...

work page 2019

[2] [2]

and Med-PaLM 2 (Singhal et al., 2023), exhibit remarkable capabilities in contextual reasoning and domain-specific tasks. These developments drive adoption across healthcare (Esteva et al., 2019; Topol, 2019), legal services (Alarie, Niblett, & Yoon, 2018; McGinnis & Pearce, 2014), financial analysis (Cao, 2022; López de Prado, 2018), and creative industr...

work page 2023

[3] [3]

As professionals refine these tools, they externalize relational implicit knowledge (Collins, 2010), reshaping the organization of knowledge work

illustrate AI’s growing reach while McKinsey (2025a) estimates a $4.4 trillion productivity boost from AI use cases, underscoring its economic impact. As professionals refine these tools, they externalize relational implicit knowledge (Collins, 2010), reshaping the organization of knowledge work. Drawing on Christensen’s (1997) disruptive innovation theor...

work page 2010

[4] [4]

Understanding this dynamic is vital for professionals, organizations, and policymakers navigating AI’s integration into knowledge-intensive fields

forecasts 170 million new jobs and 92 million displaced by 2030, netting 78 million jobs, further illustrating AI’s dual impact. Understanding this dynamic is vital for professionals, organizations, and policymakers navigating AI’s integration into knowledge-intensive fields. In the following sections, we define professional expertise, distinguishing expl...

work page 2030

[5] [5]

we can know more than we can tell,

captured this with his observation, "we can know more than we can tell," highlighting intuitions, contextual awareness, and pattern recognition honed through experience. Collins (2010) refines this into "relational tacit knowledge" (transferable via social interaction, e.g., a clinician’s diagnostic hunch based on patient cues) and "somatic tacit knowledg...

work page 2010

[6] [6]

gut feel

or architects adjust designs mid-process (Cross, 2004). Studies confirm this across fields: physicians blend analytical and intuitive reasoning (Norman et al., 2007), while executives rely on "gut feel" for strategy (Sadler-Smith & Shefy, 2004). Traditionally, tacit knowledge develops through socially embedded mechanisms. Lave and Wenger’s (1991) "legitim...

work page 2004

[7] [7]

explicitation

or Med-PaLM 2 (Singhal et al., 2023), embedding diagnostic pattern recognition. Lawyers highlight relevant passages in legal documents to enhance search algorithms, as with GPT-4 applications (Grabmair et al., 2015; Zhong et al., 2020; OpenAI, 2023), while designers select preferred outputs from generative models like Adobe Sensei to refine aesthetic para...

work page 2023

[8] [8]

In medicine, clinical expertise externalizes through multiple AI interactions

over somatic forms. In medicine, clinical expertise externalizes through multiple AI interactions. Diagnostic systems for retinal disease (De Fauw et al., 2018), dermatology (Esteva et al., 2017), and radiology (McKinney et al.,

work page 2018

[9] [9]

leverage such inputs. Clinicians’ corrections—e.g., overriding AI suggestions in diabetic retinopathy screening (Beede et al., 2020)—refine algorithms for edge cases, while integration with electronic health records (EHRs) enables continuous learning from decisions (Rajkomar et al., 2018; Chen et al., 2020). A 2024 study found 75% of U.S. hospitals using ...

work page 2020

[10] [10]

clinical intuition

with Przegalinska et al. (2025) noting AI enhances clinical outcomes through task-technology fit, embedding relational tacit knowledge into scalable systems. These interactions embed relational tacit knowledge into scalable systems, though somatic skills like physical examination resist codification. This shift challenges medicine’s reliance on tacit expe...

work page 2025

[11] [11]

or refining GPT-4-generated contracts (Alarie et al., 2018; OpenAI,

work page 2018

[12] [12]

encode judgment about relevance and phrasing. Routine research tasks—e.g., prioritizing sources (Grabmair et al., 2015; Zhong et al., 2020)—feed search algorithms, with a 2024 ABA survey reporting 80% of firms using AI weekly for such purposes (ABA, 2024). These actions transfer relational tacit knowledge, enabling systems to mimic case-specific reasoning...

work page 2015

[13] [13]

Junior attorneys traditionally learned from seniors, funded by billable hours (Henderson, 2013; Ribstein, 2010)

faces disruption. Junior attorneys traditionally learned from seniors, funded by billable hours (Henderson, 2013; Ribstein, 2010). AI’s codification of senior judgment—e.g., contract drafting—empowers less experienced practitioners (McGinnis & Pearce, 2014; Susskind, 2013), compressing hierarchies. BLS (2025a) data show a 12% drop in paralegal jobs from 2...

work page 2013

[14] [14]

Writers editing GPT-4 text (Radford et al., 2019; OpenAI,

embed preferences into algorithms. Writers editing GPT-4 text (Radford et al., 2019; OpenAI,

work page 2019

[15] [15]

Continuous feedback loops—e.g., 70% of designers tweaking AI outputs daily (AIGA, 2024)—codify relational tacit knowledge, enabling autonomous generation

teach style and tone, with Sudowrite supporting 15,000+ authors by 2024 (Clark et al., 2018; Sudowrite, 2024). Continuous feedback loops—e.g., 70% of designers tweaking AI outputs daily (AIGA, 2024)—codify relational tacit knowledge, enabling autonomous generation. McKinsey (2025a) highlights AI’s $4.4 trillion productivity potential, partly from creative...

work page 2024

[16] [16]

or wealth managers tailoring robo-advisor plans (D’Acunto et al., 2019; Lam,

work page 2019

[17] [17]

market sense

encode market judgment. Integration into workflows—e.g., 85% of U.S. trading firms using AI by 2024 (FINRA, 2024)—facilitates learning from decisions (Agrawal et al., 2019; Hammond, 2016). BLS (2025b) projects a 9.5% growth for financial analysts from 2023–2033, indicating augmentation over displacement, while CEPR (2025) reports a 0.5-0.6% productivity b...

work page 2024

[18] [18]

de-skilling

shifts attorney focus to negotiation and advocacy, per a 2024 ABA survey noting 60% of firms reallocating hours (ABA, 2024; Susskind, 2013). Task restructuring carries dual implications for expertise. It may elevate the value of tacit knowledge resistant to automation—like interpersonal skills and ethical judgment (Deming, 2017; Frey & Osborne, 2017)—with...

work page 2024

[19] [19]

social skills premium

with McKinsey (2025a) noting $4.4 trillion in productivity gains amplifying top performers. CEPR (2025) confirms a 0.5-0.6% productivity boost in Japan, suggesting broader economic shifts (CEPR, 2025). A 2024 Bloomberg report notes 10% of financial analysts shifting to oversight roles since 2020 (Bloomberg, 2024), amplifying this trend. Professional value...

work page 2025

[20] [20]

automation paradox

but may shrink future demand (Autor, 2015; Susskind & Susskind, 2015). McKinsey (2025b) notes only 1% of firms are AI-mature, suggesting early-stage productivity gains still favor collaboration. Zuboff’s (1988) "automation paradox" pits short-term gains against long-term security, varying by career stage and specialization. A 2024 survey found 60% of mid-...

work page 2015

[21] [21]

stepping up

while leveraging somatic and contextual strengths. These strategies balance collaboration with AI against potential replacement, offering practical pathways for sustained relevance. The first response, termed "stepping up" by Davenport and Kirby (2016), shifts professionals to higher abstraction and oversight as routine tasks automate. While AI excels in ...

work page 2016

[22] [22]

explainable AI literacy

add significant value, with 65% of U.S. hospitals mandating such oversight by 2024 (HIMSS, 2024). Bailey and Barley (2020) note data scientists bridging technical and domain roles thrive, a model applicable across fields. Developing supervisory expertise demands educational shifts beyond traditional knowledge. Hoffman et al.’s (2017) "explainable AI liter...

work page 2024

[23] [23]

deliberative rationality

to sustain tacit knowledge resisting externalization. Historically, these groups nurtured expertise through social participation (Lave & Wenger, 1991), and amid AI’s rise, they can prioritize human strengths (Bailey & Barley, 2020; Zuboff, 1988). This includes Dreyfus and Dreyfus’ (1986) "deliberative rationality"—critical reflection for novel contexts—an...

work page 1991

[24] [24]

design thinking

that blend judgment with augmentation. Value shifts from information control to contextual application and trust-building (Abbott, 1988; Susskind & Susskind, 2015). For instance, 80% of lawyers in a 2024 ABA survey saw AI as enhancing, not replacing, client counsel (ABA, 2024). Schön’s (1983) "design thinking" aids problem-framing, while ethics evolve—e.g...

work page 1988

[25] [25]

teaming structures

shifts to systems while somatic and contextual expertise persists. At the organizational level, externalization necessitates rethinking knowledge management, work design, and professional development. Traditional structures assume distinct expert-novice boundaries, with expertise accruing along defined career paths (Abbott, 1988; Freidson, 1970). As AI co...

work page 1988

[26] [26]

shadow learning

counter this, fostering tacit acquisition; e.g., 65% of architects in 2024 valued peer forums for design intuition (AIA, 2024). Beane’s (2019) "shadow learning"—like Johns Hopkins’ surgical labs (Topol, 2019)—and Edmondson’s (1999) "psychological safety" ensure professionals critique AI limits, with 60% of engineers reporting safer innovation spaces by 20...

work page 2024

[27] [27]

expert systems thinking

reflects this, valuing navigation of complexity over mere processing (Bailey & Barley, 2020; Pasquale, 2020). Feasibility hinges on scalable programs, though resource disparities challenge smaller firms (Brynjolfsson et al., 2018). At the educational level, the paradox upends traditional training and credentialing. Balancing theory and practice (Eraut, 20...

work page 2020

[28] [28]

technological gatekeeping

exemplify this, integrating human-tech judgment (Davenport & Kirby, 2016). At the policy level, the paradox raises questions of regulation, labor protections, and expertise access. Regulation historically ensures quality and autonomy (Abbott, 1988; Freidson, 1970), now requiring "technological gatekeeping" (Susskind & Susskind, 2015)—defining AI autonomy ...

work page 2016

[29] [29]

Human-in-the-loop

guide high-stakes use; the EU’s 2024 AI Act mandates human review in healthcare (EU, 2024). "Human-in-the-loop" rules (Pasquale,

work page 2024

[30] [30]

New social contracts

persist, with 75% of U.S. hospitals enforcing this by 2024 (HIMSS, 2024). Labor policies must adapt to disrupted career paths. "New social contracts" (Brynjolfsson & McAfee,

work page 2024

[31] [31]

countervailing investments

share tech gains equitably, with "countervailing investments" (Acemoglu & Restrepo, 2019)—e.g., $600 million in U.S. AI training grants by 2024 (DOL, 2024)—creating opportunities. "Institutional innovations" (Autor,

work page 2019

[32] [32]

Appropriate use

like tax incentives aid transitions, though funding lags (BLS, 2024). Access policies balance AI’s democratizing potential—25% legal cost cuts since 2022 (ABA, 2024)—with quality concerns. "Appropriate use" guidelines (Susskind,

work page 2024

[33] [33]

These responses acknowledge externalization as a social shift, not just a technical one

ensure accountability, as in California’s 2024 AI liability laws (CA, 2024). These responses acknowledge externalization as a social shift, not just a technical one. Preserving human judgment, creativity, and ethics—via frameworks balancing AI and human strengths— aims for a future where professional expertise is enhanced, not diminished, despite implemen...

work page 2024

[34] [34]

On the Opportunities and Risks of Foundation Models

or lawyers refining GPT-4 (OpenAI, 2023)—in a fluid, often unintended process (Amershi et al., 2014). A 2024 BLS survey notes 70% of U.S. professionals use AI daily, amplifying this shift (BLS, 2024), with implications often unrecognized until deployment. Second, it varies by domain: medicine’s diagnostic shift with 75% of hospitals using AI-enhanced EHRs...

work page internal anchor Pith review Pith/arXiv arXiv 2023

[35] [35]

S., Shulman, L

Elstein, A. S., Shulman, L. S., & Sprafka, S. A. (1978). Medical problem solving: An analysis of clinical reasoning. Harvard University Press. Eraut, M. (2000). Non-formal learning and tacit knowledge in professional work. British Journal of Educational Psychology, 70(1), 113–136. Esteva, A., Kuprel, B., Novoa, R. A., Ko, J., Swetter, S. M., Blau, H. M., ...

work page arXiv 1978

[36] [36]

CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning

Rajkomar, A., Oren, E., Chen, K., Dai, A. M., Hajaj, N., Hardt, M., … & Dean, J. (2018). Scalable and accurate deep learning with electronic health records. npj Digital Medicine, 1(1), 1–10. Rajpurkar, P., Irvin, J., Zhu, K., Yang, B., Mehta, H., Duan, T., … & Ng, A. Y. (2017). CheXNet: Radiologist-level pneumonia detection on chest X-rays with deep learn...

work page internal anchor Pith review Pith/arXiv arXiv 2018

[37] [37]

[https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work] McKinsey (2025b)

McKinsey & Company. [https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work] McKinsey (2025b). The state of AI: How organizations are rewiring to capture value. McKinsey & Company. [https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-a...

work page 2025

[38] [38]

[https://www.weforum.org/publications/future-of-jobs-report-2025/]

World Economic Forum. [https://www.weforum.org/publications/future-of-jobs-report-2025/]

work page 2025