Human Capital: Leaky Competitive Advantage

• Human Capital as a Leaky Competitive Advantage is defined by the distinction between specific expertise (SHC) and broad cognitive abilities (GHC), where AI erodes the former and amplifies the latter.
• Empirical studies in creative tasks reveal that generative AI diminishes the return on specialist skills while significantly boosting outcomes for individuals with high general human capital.
• The evolving competitive landscape prompts organizations to shift focus from narrow specialization to developing critical thinking and adaptability to sustain AI-augmented advantage.

Human capital has long been recognized as a central determinant of competitive advantage in knowledge-intensive and creative industries. Traditionally, specific human capital (SHC)—deep, domain-focused expertise—conferred strong and durable occupational moats. However, the diffusion of generative AI within creative domains is reshaping this landscape. Emerging research integrating human capital theory with the automation-augmentation framework reveals that the once-defensive advantages of specialist skills have become porous—resulting in “leaky” competitive advantage. In parallel, general human capital (GHC), defined by broad cognitive abilities such as fluid intelligence, education, and adaptability, is increasingly amplified as a new locus of sustainable value (Huang et al., 5 Dec 2024).

1. Theoretical Foundations: Human Capital Taxonomy and AI

A revised conceptual model grounded in human capital theory distinguishes between general human capital (GHC) and specific human capital (SHC). GHC comprises transferable cognitive skills—fluid intelligence, education level, critical thinking, and adaptability. SHC consists of deep expertise accumulated through prolonged domain-specific training. The automation-augmentation perspective, particularly the “automation–augmentation paradox” (Raisch & Krakowski, 2021), clarifies AI’s dual channels of impact:

• Automation: Generative AI substitutes for the knowledge and workflows previously guarded by SHC, thus diminishing the incremental return to specialization.
• Augmentation: AI’s effectiveness is enhanced by the oversight, evaluative judgment, and integration skills embedded in GHC, thereby increasing the returns to such capacities.

Formally, creative performance ($C_i$) for individual $i$ with access to AI ($AI_i$), general capital ($G_i$), and specific capital ($S_i$) is captured as:

$$C_i = \alpha + \beta_1 \cdot AI_i + \beta_2 \cdot G_i + \beta_3 \cdot S_i + \beta_4 \cdot (AI_i \times G_i) + \beta_5 \cdot (AI_i \times S_i) + \epsilon_i$$

where

• $\partial^2C / \partial AI \partial G > 0$ (augmentation effect)
• $\partial^2C / \partial AI \partial S < 0$ (automation effect)

2. Empirical Evidence from Creative Tasks

Research on generative AI’s interaction with human capital employs randomized controlled experiments in flash-fiction writing and song lyric composition. In flash fiction (N = 162), participants randomly assigned to AI assistance (GPT-4) or control conditions were evaluated on creativity (novelty, usefulness, enjoyment) with robust interrater reliability (ICC₂ > .87). GHC was assessed via Raven’s Progressive Matrices IQ scores and self-reported education; SHC used self-rated literary ability. In song lyric composition (N = 299), GHC and SHC were similarly measured, with additional objective publication experience as SHC and evaluation by raters on both lyrics alone and produced songs (ICC₂ ≈ .35–.90).

3. Channel-Specific Returns to Human Capital

The empirical findings consistently reveal divergence between GHC and SHC in the context of AI augmentation:

• General Human Capital (GHC): AI use significantly enhanced creative output for individuals with higher GHC (e.g., Experiment 1: AI × Education on Novelty β = 0.480, p = .015; AI × IQ on Novelty β = 0.193, p = .008; see also Experiment 2).
• Specific Human Capital (SHC): The incremental advantage of SHC was diminished by AI assistance (e.g., Experiment 1: Writing-skill × AI on Usefulness β = –0.600, p = .003; Experiment 2: Publication-experience × AI on song Novelty_S β = –0.221, p = .029).

The implication is that AI augments those able to critically evaluate, integrate, and direct its outputs, while automating the knowledge patterns that previously protected specialists.

4. The Concept of “Leaky” Competitive Advantage

The observed reduction in the value of SHC in the presence of generative AI constitutes a “leaky” competitive advantage. AI’s automation channel replicates or approximates domain-specific patterns, eroding the previously impermeable moat around deep expertise (β₅ < 0 in both experiments). Meanwhile, the augmentation channel channels higher returns to GHC, which cannot be easily codified or automated. This structural shift renders specialist skill less distinctive, while broad cognitive adaptability emerges as the new source of defensible organizational and individual advantage.

Human Capital Type	Impact of AI	Empirical Pattern
SHC	Automated	β₅ < 0 (leaky advantage)
GHC	Augmented	β₄ > 0 (reinforced advantage)

5. Organizational and Workforce Implications

The porous nature of SHC advantage in the AI era alters human resource management, organizational structure, and inequality dynamics:

• Talent Management: A shift away from narrowly specialized training toward broad cognitive skill development (critical thinking, learning agility) is warranted.
• Role Design: Emphasis on oversight, evaluation, and AI integration supersedes traditional specialist roles.
• Workforce Inequality: Lower barriers for domain tasks democratize access, but amplify stratification by GHC. Individuals lacking strong general cognitive capital may be less able to leverage AI, intensifying preexisting divides.
• Hierarchy and Structure: Specialist pipelines and apprenticeships may be devalued in favor of cross-functional, cognitively diverse teams. New professional hierarchies grant primacy to “AI integrators” possessing high GHC, at the expense of traditional domain veterans.

6. Synthesis and Theoretical Advancement

The contingent human capital theory advanced by integrating taxonomy with automation–augmentation logic provides a nuanced explanation for AI’s impact on skill-based advantage. As generative AI erodes the insulation once provided by deeply specialized expertise, it transforms the competitive landscape. The pebbled fortress of SHC becomes permeable under automation pressure, making competitive advantage “leaky.” In contrast, broad cognitive agility—embodied by GHC—proves resilient and is amplified by augmentation. Sustainable advantage thus resides increasingly in the ability to guide, critique, and refine AI’s outputs, rather than in the possession of niche, specialist knowledge (Huang et al., 5 Dec 2024).