Table 3 Barriers to dynamic capabilities... | Emerald Publishing

Table 3

Barriers to dynamic capabilities and AI at the individual and organisational levels

Dynamic capability category	Individual level	Interaction mechanism	Organisational level
Sensing Second-order themes First-order informant terms	Uncertainty: A lack of skills and confidence - Uncertainty and concern about potential errors related to AI use - Staff’s low competence and scepticism - A lack of AI expertise or motivation in the organisation	Bottom-up: Insufficient training and facilitating support causing confusion - Insufficient training and guidance in the organisation - Inconsistent use of AI, causing confusion and security risks Top-down: Bias towards internal knowledge stocks and lack of external expertise - Caution and regulatory-induced slowness - Resources focused on pre-studies, not on actual experimentation - Uncertainty and insufficient knowledge about regulation and application - Limited use of external AI experts, partly due to their insufficient understanding of the public sector’s needs	Lack of strategic alignment and resource allocation - Low managerial competence and critical attitude - Overly high expectations of technology that is still in development - Organisational unclarity about the short- and long-term benefits of AI and potential use cases - Organisational restrictions on the use of open tools - Low legal expertise and scepticism
Seizing Second-order themes First-order informant terms	Lack of idea generation and experimentation - Lack of AI expertise and skills - Fear of job continuity due to AI automation - Black box feature of GenAI excludes innovativeness and experimentation	Bottom-up: Uncollaborative planning and pilot programmes - Non-inclusive pilot projects that exclude end-user involvement - Non-inclusive planning of potential pilot cases Top-down: Public sector–specific external barriers to AI adoption - A shortage of AI professionals in the market - A lack of tailored and ready-made solutions for the public sector in the market - Limited support for local languages in AI language models - Privacy and security restrictions regarding internal and confidential data	Lack of strategy, scalability, and organisational leadership - The lack of a strategy, vision or action plan for AI and data - The lack of an experimentation culture - Internal overregulation may restrict innovation in the piloting phase - Without strategic alignment and proper expectation management, pilots risk remaining isolated and unscalable - A lack of support for privacy and security issues in the public sector
Trans-forming Second-order themes First-order informant terms	Lack of practical guidance and practices - A lack of established good practices for successful transformation - A lack of practical guidance for public procurement experts in tendering AI solutions	Bottom-up: Selection of an unsuitable procurement model and/or excessive requirements in AI tender processes - Broad involvement in specification work can lead to excessive requirements in the tendering process - A failed AI tender process can discourage future actions Top-down: Data availability, accessibility, and implementation model failures - A lack of machine-readable interfaces in existing systems - Bureaucracy and regulation may limit data access and analytics - Privacy and IT restrictions slow down transformation processes - Barriers and regulations restricting free movement of data between the public sector and external actors - Undervaluation and uncertainty of data quality and processing - A partnership model centred around a single LLM or provider or do-it-yourself model may prove insufficient in transformation	Challenges in change management and long-term commitment - Hierarchical organisational culture challenges - Inflexible organisational structures and processes for integrating AI capabilities into transformation - Challenges in planning and resourcing the post-pilot investment, maintenance and development phases - AI infrastructure limitations in handling larger datasets

Dynamic capability category	Individual level	Interaction mechanism	Organisational level
SensingSecond-order themesFirst-order informant terms	Uncertainty: A lack of skills and confidence- Uncertainty and concern about potential errors related to AI use - Staff’s low competence and scepticism - A lack of AI expertise or motivation in the organisation	Bottom-up: Insufficient training and facilitating support causing confusion- Insufficient training and guidance in the organisation - Inconsistent use of AI, causing confusion and security risks Top-down: Bias towards internal knowledge stocks and lack of external expertise- Caution and regulatory-induced slowness - Resources focused on pre-studies, not on actual experimentation - Uncertainty and insufficient knowledge about regulation and application - Limited use of external AI experts, partly due to their insufficient understanding of the public sector’s needs	Lack of strategic alignment and resource allocation- Low managerial competence and critical attitude - Overly high expectations of technology that is still in development - Organisational unclarity about the short- and long-term benefits of AI and potential use cases - Organisational restrictions on the use of open tools - Low legal expertise and scepticism
SeizingSecond-order themesFirst-order informant terms	Lack of idea generation and experimentation- Lack of AI expertise and skills - Fear of job continuity due to AI automation - Black box feature of GenAI excludes innovativeness and experimentation	Bottom-up: Uncollaborative planning and pilot programmes- Non-inclusive pilot projects that exclude end-user involvement - Non-inclusive planning of potential pilot cases Top-down: Public sector–specific external barriers to AI adoption- A shortage of AI professionals in the market - A lack of tailored and ready-made solutions for the public sector in the market - Limited support for local languages in AI language models - Privacy and security restrictions regarding internal and confidential data	Lack of strategy, scalability, and organisational leadership - The lack of a strategy, vision or action plan for AI and data - The lack of an experimentation culture - Internal overregulation may restrict innovation in the piloting phase - Without strategic alignment and proper expectation management, pilots risk remaining isolated and unscalable - A lack of support for privacy and security issues in the public sector
Trans-formingSecond-order themesFirst-order informant terms	Lack of practical guidance and practices- A lack of established good practices for successful transformation - A lack of practical guidance for public procurement experts in tendering AI solutions	Bottom-up: Selection of an unsuitable procurement model and/or excessive requirements in AI tender processes- Broad involvement in specification work can lead to excessive requirements in the tendering process - A failed AI tender process can discourage future actions Top-down: Data availability, accessibility, and implementation model failures- A lack of machine-readable interfaces in existing systems - Bureaucracy and regulation may limit data access and analytics - Privacy and IT restrictions slow down transformation processes - Barriers and regulations restricting free movement of data between the public sector and external actors - Undervaluation and uncertainty of data quality and processing - A partnership model centred around a single LLM or provider or do-it-yourself model may prove insufficient in transformation	Challenges in change management and long-term commitment- Hierarchical organisational culture challenges - Inflexible organisational structures and processes for integrating AI capabilities into transformation - Challenges in planning and resourcing the post-pilot investment, maintenance and development phases - AI infrastructure limitations in handling larger datasets

Source(s): Authors' own work