AI utvecklingsarbete värde kräver simultaneous orchestration av culture, competence, structure visar Wiley R&D Management-studie (291 survey + kvalitativa intervjuer): Cultural readiness driver adoption (β=0.289, p<0.01), structural readiness direkteffekt NOT significant (β=0.132, p=0.244), men här kritiskt – AI literacy modererar relationships kraftigt. Low literacy employees: structure HELPS (substitutes individual knowledge), high literacy employees: samma structure HINDERS (perceived som rigid, reduces motivation, β=-0.297 moderation). För dig som projektledare betyder detta: behandla inte kultur/struktur/kompetens som sekventiella steg. They interact dynamically. GenAI use → R&D impact (β=0.493), R&D → innovation capacity (β=0.413), men utan aligned readiness dimensions + appropriate literacy levels = suboptimal realization. Separation av dimensions = guaranteed failure.
Tre dimensioner måste fungera samtidigt, inte sekventiellt
Study design (robust):
- 291 respondents (DACH region), 50.5% male, 48.5% female
- Industries: IT/telecom (14.9%), automotive (11.4%), healthcare (8%), machinery (6.5%)
- 54.2% large corporations (>250 employees)
- Qualitative pre-study: 5 expert interviews (R&D managers, product developers, startup founders)
- SEM analysis med structural equation modeling
Three readiness dimensions tested:
1. Structural readiness (technological/procedural):
- AI-compatible work practices
- Technical tools & infrastructure
- Data access för AI applications
2. Cultural readiness (psychological/social):
- Openness to AI-driven change
- Clear implementation goals
- AI-compatible corporate values
- Appropriate training measures
3. AI literacy (individual competence):
- Basic AI knowledge
- Algorithm understanding
- Practical experience med AI
För projektledare: Traditional approach = fix structure first, then culture, then train people. WRONG. These dimensions interact. Optimizing one without others = waste eller backfire.
Cultural readiness: Starkt direkt impact, konsistent across literacy levels
Cultural readiness → GenAI utilization:
- β=0.289 (p<0.01) – second strongest direct effect i modellen
- Moderation av AI literacy: NOT significant (β=0.094, p=0.241)
Meaning: Kultur works UNIFORMLY across all employee literacy levels. Open culture med clear goals, AI-compatible values, training support enables adoption whether employee har low eller high AI understanding.
Practical markers av cultural readiness:
- Leadership frames AI som opportunity, not threat
- Clear communication om implementation goals
- Values alignment (experimentation encouraged, failure tolerated)
- Systematic training provisions
- Data sensitivity balanced med innovation needs
För projektledare: Culture = universal enabler. Unlike structure (som varies i effectiveness beroende på employee literacy), positive AI culture benefits everyone equally. This is WHERE to invest FIRST if forced to prioritize.
Structural readiness paradox: Helps novices, hinders experts
Structural readiness → GenAI utilization:
- Direct effect: β=0.132 (NOT significant, p=0.244)
- Moderation av AI literacy: β=-0.297 (HIGHLY significant, p<0.01)
Simple slope analysis reveals:
Low AI literacy (-1 SD):
- Structural readiness STRONGLY POSITIVE impact
- Steep positive gradient
- Structure substitutes för limited individual knowledge
- Provides guidance, lowers barriers
- Employees need procedural support
High AI literacy (+1 SD):
- Structural readiness becomes NEGATIVE
- Negative gradient
- Same structures perceived som constraints
- Reduces motivation, hinders experimentation
- Employees want flexibility, not rigidity
För projektledare: One-size-fits-all structural approach = guaranteed suboptimal outcomes. Design tiered structural frameworks:
- Novices: Detailed procedures, step-by-step guides, restricted tool access (safety rails)
- Experts: Flexible protocols, broader permissions, experimentation space
GenAI utilization → R&D impact: Strong men literacy-moderated
GenAI use → R&D activities impact:
- β=0.493 (p<0.001) – STRONGEST effect i entire model
- Supports: idea generation, material research, prototyping, topology optimization, product alternatives
- Reduces development timelines significantly
Men moderation av AI literacy:
- β=-0.135 (p<0.05) – negative moderation (unexpected!)
Simple slope analysis: Low AI literacy:
- Steeper positive gradient
- Larger perceived immediate benefits
- Efficiency gains highly visible
High AI literacy:
- Flatter gradient (still positive but weaker)
- More critical evaluation av AI outputs
- Higher expectations reduce perceived incremental value
- Cautious integration into workflows
Explanation via absorptive capacity theory: High AI literacy = stronger potential absorptive capacity (acquire/assimilate knowledge) MEN simultaneously more selective transformation/exploitation process. Critical filtering reduces apparent short-term impact even though long-term integration may be more sophisticated.
För projektledare: Paradoxically, high-literacy employees report LESS dramatic R&D improvements trots likely using AI more effectively. This is measurement artifact, not actual performance. Track actual outcomes (prototypes created, development time, product alternatives generated), inte bara perceived impact surveys.
R&D impact → Innovation capacity: Robust effect
R&D improvements → Innovation capacity:
- β=0.413 (p<0.001) – second strongest effect
- NOT moderated av AI literacy (β=0.045, not significant)
- Uniform impact across employee competence levels
Innovation capacity operationalized:
- More efficient development processes
- Faster product development timelines
- Increased product innovation output
För projektledare: Once R&D improvements materialized (regardless av employee literacy), translation till organizational innovation capacity är AUTOMATIC. This validates GenAI som legitimate innovation driver, not just efficiency tool.
Fem konkreta orchestration strategies
1. Simultaneous dimension assessment: Before GenAI rollout: measure all three simultaneously
- Cultural readiness survey (openness, values, goals)
- Structural audit (infrastructure, processes, data access)
- AI literacy baseline (knowledge, experience, confidence)
Identify misalignments BEFORE implementation
2. Differentiated structural scaffolding: Low literacy cohort: Detailed procedures, guided workflows, constrained tool access High literacy cohort: Flexible protocols, broad permissions, experimentation budgets Avoid one-size-fits-all rigidity
3. Universal cultural investment: Since culture works equally across literacy levels → invest heavily Leadership messaging, value alignment, goal clarity, training programs This dimension has NO downside regardless av employee competence
4. Literacy development programs med different goals: Novices: Foundation building (what is AI, how use safely, basic prompting) Intermediates: Application focus (R&D-specific use cases, workflow integration) Experts: Advanced optimization (custom configurations, critical evaluation, experimentation design)
5. Measurement strategy accounting för moderation effects: Track actual R&D outcomes (prototypes, timelines, product alternatives) NOT just perceived impact surveys (biased av literacy levels) High-literacy employees will report smaller gains despite potentially stronger actual performance
Absorptive capacity & dynamic capabilities framing
Study explicitly positions findings via:
Absorptive capacity (Cohen & Levinthal 1990): Organization’s ability recognize value av new external information, assimilate, apply till commercial ends
- Potential AC: Acquire + assimilate knowledge
- Realized AC: Transform + exploit knowledge
Structural/cultural readiness = organizational-level microfoundations AI literacy = individual-level microfoundation
Critical insight: AC NOT simply strengthened av accumulation av både organizational + individual microfoundations. Depends på alignment mellan them.
Low literacy: Structural readiness strengthens AC genom compensating individual limitations High literacy: Samma structural mechanisms kan restrict transformation/exploitation, dampening realized AC
Dynamic capabilities (Teece 2007):
- Sense opportunities (requires employee knowledge)
- Seize opportunities (requires organizational readiness)
- Reconfigure assets (requires literacy + readiness alignment)
För projektledare: Theoretical implication = PRACTICAL insight. You cannot optimize organization’s AI absorptive capacity genom just adding more training ELLER just improving infrastructure. Misalignment between levels REDUCES total capacity even when both individually strong.
Bottom line
AI utvecklingsarbete värde kräver simultaneous orchestration culture, structure, competence – separation = failure. Wiley study (291 survey) visar cultural readiness strong universal enabler (β=0.289, uniform across literacy), structural readiness paradoxical (helps novices, hinders experts via β=-0.297 moderation), GenAI use → R&D impact strong (β=0.493) men literacy-moderated (high literacy reports smaller gains despite likely better actual use). R&D → innovation capacity robust (β=0.413, not moderated). Practical implications: (1) assess three dimensions simultaneously pre-rollout, (2) differentiated structural scaffolding per literacy cohort, (3) universal cultural investment (no downside), (4) literacy programs med cohort-specific goals, (5) measure actual outcomes not perceived impact. Theoretical framing via absorptive capacity: alignment between org-level (readiness) + individual-level (literacy) microfoundations critical – accumulation alone insufficient. Dynamic capabilities require sensing (literacy) + seizing (readiness) coordination.
Källa: “Organizational Readiness, AI Literacy, and the New Frontier of R&D: How Generative AI Shapes Innovation Capacity” av Stefanie Steinhauser & Paulina Heid, Technical University of Applied Sciences Amberg-Weiden, publicerad mars 2026 (R&D Management journal).