Toward AI Governance: Identifying Best Practices and Potential Barriers and Outcomes

The quote only partly states this claim.

AI was found to be most helpful for improving confidence in the results and final products.

“Among the outcomes of the study, AI was found to be most helpful for (1) reducing maintenance costs, (2) increasing flexibility and robustness of the development process, (3) improving confidence in the results and final products, and (4) gaining a competitive edge over the competition.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Legal regulations restrict the cases to using AI in specific areas, for instance they are not allowed to speculate on prices.

“while legal regulations restrict them to using AI in specific areas, for instance they are not allowed to speculate on prices.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Company A collected necessary data through vendors, enabling comparison of data and ensuring high quality outcomes for trained models.

“The necessary data was collected through vendors, so it was possible to compare data and ensure high quality outcomes for the trained models.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Domain experts participated in evaluation in Company A to provide insights and feedback to make model outputs reliable and trustworthy.

“Also, domain experts participated in the evaluation so they can provide their insights and feedback to make the model outputs reliable and trustworthy.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Control of key domain knowledge was one of Company A's main concerns; it did not want to give away domain knowledge to external partners because they could build and sell similar AI products to rivals.

“Control of key domain knowledge was one of the main concerns for firm. Company A did not want to give away domain knowledge to external partners, who offer specialized AI products, since they could build and sell similar AI products to their rivals”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company A's AI team brought value by forecasting energy consumption, assisting in decision-making for end-users, and automating repetitive tasks.

“The AI team brought value to the firm by forecasting energy consumption, assisting in decision-making for the end-users and automating repetitive tasks.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company A replaced Excel file data exchange with APIs because Excel files were not secure and could not scale up.

“In the beginning, data was exchanged through Excel files. These files were not secure, and at the same time they realized that they could not scale up, so APIs were used to replace Excel files.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

The set-up goals of the cases mainly concern reducing costs and forecasting energy demands.

“Lastly, the set-up goals mainly concern reducing costs and forecasting energy demands.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

All cases operate in the same industry and have overlapping areas of operation.

“Firstly, all the cases operate in the same industry and have overlapping areas of operation.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

As a result of the AI team's work, Company A's performance was boosted and maintenance costs were down.

“As a result, performance was boosted and maintenance costs were down.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Company A developed intelligence that runs on top of cloud services to achieve automation and flexibility while avoiding developing the entire software from scratch.

“The development team aimed for automation and flexibility but they did not want to develop the entire software from scratch since it would be time-consuming to do everything. At the same time, they did not prefer to use software of other companies, so they decided to develop the intelligence that runs on top of cloud services”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company A created ML pipelines in the cloud to govern data cleaning and model evaluation, making it easier to oversee the overall process and apply quantifiable metrics on ML results.

“To govern the process of data cleaning and model evaluation, ML pipelines were created in the cloud. This made it easier to oversee the overall process and apply quantifiable metrics on the ML results.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Company A maintained local ML pipelines to support decision-making in case of failure, ensuring a reliable and robust system that could always generate output.

“In case of failure, local systems (ML pipelines) were ready to support decision-making, ensuring a reliable and robust system that could always generate output and assists employees with their everyday tasks”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

The selected firms share the perspective that AI must be developed, expanded and adopted in the following years as it will be crucial for gaining or maintaining competitive advantage.

“The most common perspective among the selected firms is that AI must be developed, expanded and adopted in the following years as it will be crucial for gaining or maintaining their competitive advantage over rivals or new companies entering the frame and seeking a piece of the pie.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Development best practices were followed in all cases, such as the use of Git, documentation, and containerization platforms like Docker.

“Secondly, development best practices were followed such as the use of Git, documentation and containerization platforms like Docker.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

AI contributes to the reduction of predictive maintenance costs in Company A, where maintenance is challenging due to harsh Norwegian weather, especially during winter.

“AI contributes to the reduction of predictive maintenance costs, which is challenging in Norway due to its harsh weather conditions, especially during winter.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company A restricted data access to intranet to increase security, but did not define clear data management roles, making the data request process time-consuming.

“To increase security, data access was only possible through intranet, but the company did not define clear data management roles, making the data request process time consuming”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company A prioritized prediction safety over profitability, emphasizing reducing errors rather than increasing profit margins.

“Rather than increase profit margins, the model outputs emphasize reducing errors, because Company A places higher priority on prediction safety instead of profitability.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

The quote only partly states this claim.

AI was found to be most helpful for gaining a competitive edge over the competition.

“Among the outcomes of the study, AI was found to be most helpful for (1) reducing maintenance costs, (2) increasing flexibility and robustness of the development process, (3) improving confidence in the results and final products, and (4) gaining a competitive edge over the competition.”

May combine two claims.

Company A's team decided through internal workshops to unify used tools (e.g., programming languages, databases) and create a shared vocabulary through collaborative wiki pages.

“The team decided through internal workshops to unify the used tools (e.g., programming languages, databases) while creating a shared vocabulary through collaborative wiki pages”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company B gathered data internally and externally from various vendors to verify and ensure data quality for AI models.

“To build the AI capabilities data were gathered internally and externally from various vendors as it needed to verify and ensure the quality of the data since it is crucial for the AI models”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Company C created procedures covering any AI behavior change and when timeline changes are allowed to be published in production to avoid AI failures.

“To ensure robustness, the development team has created procedures that covers extensively any AI behavior change and when the timeline that these changes are allowed to be published, for example, not before a big event, in production to avoid AI failures.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

The obstacles identified during the process of deploying AI are observable at different phases and concern different job roles.

“we find that the obstacles that are identified during the process of deploying AI are observable at different phases and concern different job roles.”

May combine two claims.

Reducing maintenance costs and forecasting energy consumption were the top priorities in the case firms, since most of the business value comes from these two outcomes.

“The need for reducing maintenance costs and forecasting energy consumption were the top priorities since most of the business value come from these two outcomes”

Probably not a substantial claim: may not pass S3 (load-bearing).

Radical changes in processes and operations made AI training for end-users more than necessary in Company A.

“Due to radical changes in processes and operations AI training for end-users was more than necessary.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

With help of AI developers, domain experts built notification systems by declaring which notifications should be sent via email and which displayed in dashboards.

“Also, with the help of AI developers they built notifications systems, by declaring which notifications should be sent via email and which should be displayed in dashboards”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

By including notification systems, error detection, and decision-making tools, firms measure the credibility of their systems and evaluate performance through KPIs.

“In addition, intelligence systems should include notification systems, error detection and decision-making tools. By doing so, firms measure the credibility of their systems and evaluate the performance gained through KPIs”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

May combine two claims.

Company A tested models against real-time datasets after deployment, enabling adjustments, better understanding of data, and improved overall quality of the system.

“Another way to ensure robust outcomes after deployment was to test the models against real-time datasets. Through This, they were able to make adjustments to the models, obtain a better understanding of the data, and improve the overall quality of the system”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Domain experts managed projects in Company A because their knowledge and expertise were needed at each step of development; their insights could determine which data were needed for ML models.

“Domain experts manage the projects as their knowledge and expertise are needed at each step of the development phase. For example, their insights could determine, which data should be needed for the machine learning models.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

In all cases, domain experts were involved in all development phases because their domain knowledge was crucial to project success and they led projects as project managers.

“In all cases, domain experts were involved in all development phases for two reasons. Firstly, their domain knowledge was crucial to the success of the project, and secondly, they led the projects as project managers.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company B developed AI models in conjunction with domain experts.

“they have begun developing AI models in conjunction with domain experts.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

It is common in AI projects to reuse parts of one pipeline for another, which reduces overall development time considerably.

“It is common in AI projects to reuse parts of one pipeline for another, which reduces the overall development time considerably.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company B reduced a cost of around two million a year to around ten thousand a year.

“We had a cost of around two million a year and it has been reduced to around ten thousand a year, almost nothing”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company B kept system complexity low to prevent high development costs while introducing end-users to AI capabilities to enable trust and adoption.

“The complexity of the system was kept low to prevent high development costs while the end-users were introduced to AI capabilities to enable them to trust and adopt AI in their daily work.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company B adopted Docker from the start to package applications into containers, and these standardized executable components boosted flexibility.

“By contrast, a containerization platform like Docker was adopted from the start to let developers to package applications into containers. Thus, these standardized executable components boosted flexibility and the cloud services were put aside.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company C has AI unit tests in place to ensure system outputs are reliable.

“AI unit tests are also in place to ensure the system's outputs are reliable.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Data management practices in the cases mostly consider securing data, using secure databases and intranet access, and creating a few roles for data access, usually two types: developers with full access and end-users with access to specific data.

“Data management practices consider mostly securing data, using secure databases and intranet access, and creating a few roles for data access, where usually there are two types of roles, (1) developers with full access and (2) end-users with access to specific data”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

May combine two claims.

Company C never experienced problems related to AI fear because all employees have a good understanding of what AI can offer and how it helps them.

“It is worth mentioning that Company C never experienced any problems related to AI fear since all employees have a good understanding of what AI can offer and how it helps them in their everyday lives.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Managers need to allocate the right resources and plan precisely because costs and timelines for AI projects do not follow the usual software projects.

“Also, managers need to allocate the right resources and plan precisely as the costs and timelines for AI projects do not follow the usual software projects”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

May combine two claims.

End-users had to follow AI suggestions intuitively and use their domain knowledge to fill gaps that AI was not capable of.

“Nevertheless, the end-users had to follow the AI suggestions intuitively and use their domain knowledge to fill gaps that AI was not capable of.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

May combine two claims.

All firms aimed to build new capabilities using external software; algorithms, trading strategies, and ML pipelines are developed by internal AI teams using platforms from third partners, keeping domain knowledge in house.

“As far as the procedural practices are concerned, all firms aimed to build new capabilities using external software. Algorithms, trading strategies and machine learning pipelines are developed by the internal AI teams, using platforms from third partners, keeping domain knowledge in house.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

May combine two claims.

Company B uses AI as a tool for prediction for identifying market opportunities and reducing maintenance costs.

“To sum up, company B uses AI as a tool for prediction for identifying market opportunities and reducing maintenance costs.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Despite use of ML applications with neural networks, all Company B's applications are considered weak AI (limited to a narrow task).

“Despite the use of ML applications with neural networks, all the applications are considered to be weak AI (AI that is limited to a narrow task).”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

In Company B, developers spent many hours explaining what a statistical output is, how the model makes predictions, and what is possible and not doable; training lasted for months.

“the developers spent many hours explaining what a statistical output is and how the model actually make predictions. Furthermore, they elaborated on what is possible and what is not doable, which took a long time for the end-users to digest all these new information and the training process lasted for months.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

When people start using the applications in Company B, they misunderstand AI capabilities; end-users had unrealistic expectations of what the model could or should predict.

“Nevertheless, when people start using the applications, they misunderstand the AI capabilities. End-users had unrealistic expectations of what the model could or should predict”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company C formed a dedicated team to deal with privacy issues by introducing procedures during data transformation and storage phases.

“As previously mentioned, the firm uses private data, so a dedicated team was formed to deal with privacy issues by introducing procedures during the data transformation and data storage phases”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Pipelines provide confidence in the quality of the end result as the final product is robust, easily maintainable, and extendable for new features.

“At the same time, pipelines provide confidence in the quality of the end result as the final product is robust, easily maintainable and extendable for new features.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

It is expensive to add many features and takes a lot of time to develop in Company B.

“Nevertheless, it is expensive to add many features and takes a lot of time to develop.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Reducing maintenance costs and errors, and creating flexible systems that can scale, were top priorities for Company B.

“Reducing maintenance costs and errors, as well as creating flexible systems that can scale, were all top priorities.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

External AI consultants assisted only at the beginning and were called on in rare cases when the development team was unsure how to proceed with a particular project.

“External AI consultants assisted only at the beginning, and they were only called on in rare cases when the development team was unsure how to proceed with a particular project”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

May combine two claims.

In Company A, changes caused human agents to feel phobic when interacting with the machine and raised concerns about being automated away and unemployed.

“All these changes caused human agents to feel phobic when interacting with the machine, as they had the overall watch and check periodically that everything is in working order. From the employees’ point of view, these automations raised concerns as they saw themselves being automated and driven away from their posts, which could result in unemployment”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company C uses cloud services to cover areas where development team members have no expertise or time to develop.

“In addition, cloud services are used to cover areas that the members of the development team have no expertise or the time to develop”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company C's growth led to introducing data management roles and restrictions on data types and situations under which employees can access data, accomplished through data-gates.

“The growth in numbers led to the decision of introducing data management roles and restrictions on the data types and situations under which employees can access data. This was accomplished through data-gates where employees had to ask for permission from the supervisor of the system”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company B decided not to have a dedicated employee responsible for data management because of the simplicity in roles and because they do not deal with private data in their applications.

“This simplicity in roles and the fact that they do not deal with private data in their applications led to the decision to not have a dedicated employee responsible for data management.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Because of the weak AI limitation, Company B still uses conventional and traditional ways in parallel with AI, while planning to replace them over time.

“Because of that the company still uses conventional and traditional ways in parallel with AI, while they plan to replace them over time in the future”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Company C's AI applications focus on specific needs, usually involving forecasting ancillary services, customer needs (AI assistants), and reducing maintenance costs by minimizing business risks.

“The AI applications focus on specific needs, which usually involve forecasting ancillary services, customer needs (AI assistants) and reducing maintenance costs by minimizing business risks at the same time.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Two main reasons Company C did not experience AI fear are extensive onboarding training and that job applicants are aware the firm uses extensive AI products and have prior knowledge or willingness to embrace AI.

“Two could be the main reasons for that. Firstly, employees have an extensive onboarding training process and secondly, people who applied to the company are aware that the firm uses extensive AI products; thus, work candidates have prior knowledge of AI technology and AI products or are willing to embrace AI.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company C uses APIs from different vendors to gather data for AI models.

“To gather the data for their AI models, Company C uses APIs from different vendors.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Customer feedback drove Company C to become more efficient and build new highly demanded services.

“A direct effect was that customers came with constructive feedback driving the firm to become even more efficient and building new services that were highly demanded”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company A held regular meetings explaining what AI can and cannot do to address employee fears.

“so we had regular meetings explaining what AI can do and not”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Company B initially used cloud services but moved to InfluxDB because cloud services were not flexible enough.

“Initially, the team used cloud services, but they were not flexible enough, or at least to their liking, so they moved to influx databases that allow storing and retrieving time series data.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

Company C identified a big gap in the market because energy-producing companies did not offer customized services.

“Company C realized that there was a big gap in the market since energy-producing companies did not offer any customized services.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Legal regulations forbid certain uses of AI, for example the prediction of energy prices.

“On top of that legal regulations forbid certain uses of AI, for example, the prediction of energy prices.”

May combine two claims.

Employees in the case firms did not trust the AI outcomes and sometimes described the recommendations as naive.

“Employees did not trust the outcomes, sometimes they described the recommendations as naive, and most importantly, employees saw AI and automation as a way of losing their status and position.”

May combine two claims.

Flexibility, productivity, and reduction of costs were immediate positive effects that Company A saw, allowing them to remain competitive and achieve their AI strategy goals.

“Flexibility, productivity, and reduction of costs were the immediate effects that the company saw as positive results allowing them to remain a competitive player while achieving their set up goals of their overall AI strategy.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

May combine two claims.

Domain experts helped create meaningful dashboards for alerting information to employees, explaining historical data, and assisting in decision making for end users.

“In addition, domain experts help with the creation of meaningful dashboards that are responsible for alerting information to employees, explaining historical data and assisting in decision making for end users.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

It became essential to increase the speed of models and scale up because the company increased the amount of data while creating new intelligence based on the data.

“Furthermore, it became essential to increase the speed of models and scale up because the company increased the amount of data while creating new intelligence based on the data.”

May combine two claims. The quote only partly states this claim.

Company B used sensors and autoregressive (AR) models to detect anomalies in the system, saving time and effort and reducing maintenance costs in the long run.

“Based on these inputs, autoregressive (AR) models were developed to detect anomalies in the system, saving time and effort, which means fewer maintenance costs in the long run”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The changes toward cloud and automation boosted efficiency, and employees liked automation that lifted the heavy load of the work.

“These changes were boosting efficiency and employees liked automation that lifted the heavy load of the work”

May combine two claims.

Companies in this study tried to create ML pipelines and interactive dashboards, but not all of them had a real focus on explainability of the results since they are still in early stages and focus on parts they believe are more urgent.

“companies in this study tried to create ML pipelines and interactive dashboards, but not all of them had a real focus on explainability of the results since they are still in early stages and focus on parts that they believe are more urgent.”

May combine two claims.

Different concerns emerge at different phases of AI projects, so AI governance also encapsulates a temporal angle in its formation and deployment.

“In addition, different concerns emerge at different phases of AI projects, so AI governance also encapsulates a temporal angle in its formation and deployment.”

Classified as probably not a substantial claim.

AI governance requires continuous adaptation and modification as new data emerges or conditions change, for instance how employees perceive AI.

“One of the main challenges we identify is that AI governance requires continuous adaptation and modification as new data emerges or conditions change, for instance how employees perceive AI.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Employees appreciate AI automating repetitive tasks since they do not want to do monotonous work.

“Firms should use AI for automating tasks that are repetitive, which is appreciated by employees since they do not want to do monotonous work”

May combine two claims.

AI governance is a multi-faceted issue for organizations and spans multiple levels, therefore requiring a structured approach when it is deployed.

“the study reveals not only that AI governance is a multi-faceted issue for organizations but that it spans multiple levels, therefore requiring a structured approach when it is deployed.”

May combine two claims.

The firms studied were leveraging AI to help them reinvent their operations, following an information analysis approach rather than an information collection approach.

“The firms we looked at were leveraging AI to help them reinvent their operations. Instead of having an information collection approach, these firms followed an information analysis approach.”

Classified as probably not a substantial claim. May combine two claims.

The interviewees are experienced and have a solid understanding of AI development over time.

“This means they are experienced, and they gained a solid understanding of AI development over time.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Audio files were transcribed verbatim, making raw data transparent and findings reproducible.

“The audio files were transcribed in a verbatim way so that the text remains identical to the audio, meaning that all raw data are transparent, and the findings and results could be reproduced and tracked down rigorously.”

May combine two claims.

AI has the potential to generate business value in performance, productivity and effectiveness, but it is not autonomous; it works in concert with human capabilities.

“While AI has the potential to generate business value in terms of performance, productivity and effectiveness, it is not autonomous, as it works in concert with human capabilities (Zhang et al., 2021).”

• cites Zhang et al., 2021 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Organizational capabilities result from combining and deploying multiple complementary resources within a firm to achieve competitive advantage.

“Consequently, organizational capabilities are the results of combining and deploying multiple complementary resources within a firm to achieve competitive advantage (Mikalef & Gupta, 2021).”

• cites Mikalef & Gupta, 2021 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Comparing the selected companies is fair because they are all in Norway, have similar AI teams in size and experience, and limited cultural differences.

“Comparing the selected companies is fair because (1) they are all allocated in Norway, (2) they have similar AI teams in terms of size and experience, although the size of the companies ranges, and (3) their cultural differences are limited.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

AI algorithms can be considered performative: they assist in decision-making, can form organizational processes, or take autonomous decisions, leading to new organizational capabilities.

“Simultaneously, the AI algorithms can be considered performative in the sense that they assist in decision-making, the extent to which their use can form organizational processes, or even take autonomous decisions (Faraj et al., 2018; Grønsund & Aanestad, 2020) that leads to new organization capabilities through AI.”

• cites Faraj et al., 2018 sourceNot checked yet

  Not checked in the free run.

• cites Grønsund & Aanestad, 2020 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

The nature of AI projects undertaken by firms indicates they face similar challenges, so they require similar solutions.

“Also, the nature of AI projects undertaken by firms indicates that they face similar challenges, so they require similar solutions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The text and audio were checked manually to ensure they matched.

“As part of the process, we had to go through the text and the audio to make sure everything was looking good since we wanted our text to match the audio and the only way to guarantee that was by checking all results manually.”

May combine two claims.

Businesses have implemented and deployed AI solutions to automate processes, increase efficiency and reduce costs, in order to gain competitive advantage.

“In order to gain a competitive advantage over their competitors (Raisch & Krakowski, 2021), businesses have implemented and deployed AI solutions to automate their processes, increase efficiency and reduce costs (Frank et al., 2019; Gregory et al., 2020).”

• cites Raisch & Krakowski, 2021 sourceNot checked yet

  Not checked in the free run.

• cites Frank et al., 2019 sourceNot checked yet

  Not checked in the free run.

• cites Gregory et al., 2020 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Dependability was achieved by being consistent in the analysis process and being in line with the accepted standards.

“Dependability was achieved by being consistent in the analysis process and being in line with the accepted standards.”

May combine two claims.

Businesses using AI continue to face challenges and are unable to immediately realize performance gains.

“However, despite the use of AI, businesses continue to face challenges and are unable to immediately realize performance gains.”

• cites Van den Broek et al., 2021Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The firms have common traits and operations, but they have some key differences in their business strategy.

“In terms of transferability, the firms have common traits and operations, but they have some key differences in their business strategy.”

May combine two claims.

There is a gap in AI governance, which deals with both IT governance and data governance, and has a direct relationship with AI.

“There is, however, a gap in AI governance, which deals with both IT governance and data governance, and has a direct relationship with AI (Mikalef et al., 2020).”

• cites Mikalef et al., 2020 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

The study enhanced research methodology by strengthening credibility, dependability, transferability and confirmability.

“Trustworthiness in the evaluation process and the findings themselves were of the utmost importance; thus, we enhanced the research methodology by strengthening credibility, dependability, transferability and confirmability (Korstjens & Moser, 2018; Sikolia et al., 2013).”

• cites Korstjens & Moser, 2018 sourceNot checked yet

  Not checked in the free run.

• cites Sikolia et al., 2013Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Firms need to introduce robust AI systems and mitigate AI risks, emphasizing the importance of creating suitable AI governance practices.

“Furthermore, firms need to introduce robust AI systems and mitigate AI risks, which emphasizes the importance of creating suitable AI governance practices.”

• Also stated in: § Governance of AI Projects · hedged

  “Allowing the development of AI applications that are not under any supervision could be harmful (Chatterjee et al., 2020; Mishra & Pani, 2020); thus, it is important to promote a trustworthy AI that is lawful (complying with laws and regulations), ethical (ensuring ethical principles and values) and robust (from a technical and social perspective).”

• cites Chatterjee et al., 2020Not checked yet

  Not checked in the free run.

• cites Mishra & Pani, 2020 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

Data were analyzed and coded independently by three authors bringing various insights and points of view, allowing identification of similarities and differences, creating a comprehensible and coherent framework.

“What is more, data were analyzed and coded independently by three authors bringing various insights and points of view so that the authors could identify similarities and differences in their results, creating a comprehensible and coherent framework.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Documents served as validations for findings.

“These documents served both as validations for our findings as well as information that we did not have prior to the interviews, assisting us to obtain a better understanding of the vision, objectives and regulations of each company.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

When a firm optimizes firm-level resources and adopts AI technological innovations, it can enhance transformed projects' business value, driving business value and impacting performance.

“When a firm optimizes its firm-level resources and adopts AI technological innovations, it can enhance its transformed projects' business value which drives business value and impacts performance (Wamba-Taguimdje et al., 2020).”

• cites Wamba-Taguimdje et al., 2020 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

All firms operate in the same industry and have similar capabilities in collecting, analyzing and interpreting data for making business decisions.

“All firms operate in the same industry and have similar capabilities in terms of collecting, analyzing and interpreting data for making business decisions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Three iterations of data analysis were necessary to develop a theory based on empirical data.

“Hence, in order to develop a theory based on empirical data, it was necessary to establish three iterations of data analysis.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The researchers validated findings by examining reports, public information, and presentations related to the research, focused on AI aspects.

“In order to obtain a high level of confidence, researchers validated findings by examining reports, public information and presentations related to this research and focused on the AI aspects (Table 3).”

Classified as probably not a substantial claim. May combine two claims.

A total of 15 individuals were interviewed, including both domain and technical experts who have worked in their current positions for at least one year, but have relative experience of at least five years.

“A total of 15 individuals were interviewed, including both domain and technical experts who have worked in their current positions for at least one year, but have relative experience of at least five years.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The value of AI in organizations varies based on the sector and the organization's activity.

“The value of AI in organizations varies based on the sector and the organization's activity(Collins et al., 2021).”

• cites Collins et al., 2021 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The similarity of outcomes across the cases could be because their desired goals were similar.

“That could be because their desired goals were similar.”

May combine two claims.

Enablers include existing AI culture and architecture within a company, whereas inhibitors are mostly legal constraints, domain challenges, high development costs, and AI-phobia.

“Enablers include existing AI culture and architecture within a company, whereas inhibitors are mostly legal constraints, domain challenges, high development costs, and AI-phobia.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Lack of domain knowledge or lack of data for creating business intelligence can be an inhibitor.

“Another inhibitor could be lack of domain knowledge or lack of data for creating business intelligence.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Building AI capabilities would be impossible for Company C if it did not follow best development practices.

“Building these AI capabilities though would be impossible if the company did not follow best development practices.”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.

Probably not a substantial claim: may not pass S3 (load-bearing).

AI development is usually more expensive than traditional software development.

“A note of caution is that AI development is usually more expensive than traditional software development”

• AssumptionRests on an assumption: The interviewees' accounts accurately reflect the firms' AI governance practices.
• AssumptionRests on an assumption: The three case companies are representative of firms in the energy sector or of AI governance best practices, so practices identified in them can be recommended for other companies.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Lack of AI understanding could lead to AI phobia, which is a huge inhibitor when a digital transformation process is in place.

“The lack of AI understanding could lead to AI phobia, which is a huge inhibitor when digital transformation process is in place.”

Classified as probably not a substantial claim. May combine two claims.

The proposed model includes structural, procedural, and relational components as key components and illustrates the techniques that companies have used over the last five years.

“Our model (Fig. 1), which includes the structural, procedural, and relational components as key components, illustrates the techniques that companies have used over the last five years.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

AI culture promotes the acceptance of AI, meaning that employees use and trust AI.

“AI culture promotes the acceptance of AI, meaning that employees use and trust AI.”

Classified as probably not a substantial claim. May combine two claims.

The model's most essential results are a competitive advantage, cost reduction, and dependable AI systems.

“The model's most essential results are a competitive advantage, cost reduction, and dependable AI systems, all of which are critical to any business's success, particularly in competitive marketplaces.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Superiority in results boosts confidence in decisions and the potential customer value is high, especially for firms that have a more direct relationship with their clients.

“This superiority in results, boosts confidence in decisions and the potential customer value is high, especially for firms that have a more direct relationship with their clients.”

Probably not a substantial claim: may not pass S2 (checkable).

Numerous impediments can lead to failure and waste of company resources.

“since numerous impediments can lead to failure and waste of company resources”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Companies gain a significant competitive advantage over the competition as they can reduce the overall product cost and provide clients with exceptional services adapted to their specific needs.

“As a result, companies gain a significant competitive advantage over the competition as they can reduce the overall product cost and provide clients with exceptional services that adopt in their specific needs.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The significance of governing AI can be critical in attaining digital innovation.

“The significance of governing AI can be critical in attaining digital innovation.”

Classified as probably not a substantial claim.

A unified system for building AI pipelines makes the system more robust, as it will be easier to maintain and improve different components of the system.

“Hence, the system will be more robust as it will be easier to maintain and improve different components of the system.”

May combine two claims.

Although several interviews were conducted with key employees within the organizations, the data collection was based on a snapshot in time and may not accurately reflect the complete breadth of practices.

“Second, while we conducted several interviews with key employees within the organizations, our data collection was based on a snapshot in time and may not accurately reflect the complete breadth of practices.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Understanding how people deal with emotional challenges is crucial for AI systems to emulate human reasoning.

“It is worth mention that emotional intelligence is not part of these systems although understanding how people deal with emotional challenges is crucial for AI systems to emulate human reasoning (Luong et al., 2021).”

• cites Luong et al., 2021 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The black-box nature of models and AI in general can be less problematic because the use of data visualizations simplifies complex data sets and provides end-users useful information that can affect business performance.

“That way the black-box nature of models and AI in general can be less problematic because the use of data visualizations simplifies complex data sets and provides end-users useful information that can affect business performance.”

May combine two claims.

The data were collected through interviews with companies that do not require extensive use of sensitive data, so there might be bias in the data or an incomplete picture of the challenges around relevant practices.

“First, the data are collected through interviews with companies that do not require extensive use of sensitive data; thus, there might be bias in our data or provide an incomplete picture of the entire challenges around relevant practices.”

Classified as probably not a substantial claim. May combine two claims.

Data visualization facilitates greater transparency and a more direct way of revising the models used to analyze data.

“This in turn facilitates greater transparency and a more direct way of revising the models used to analyze data.”

Probably not a substantial claim: may not pass S3 (load-bearing).

AI allows organizations to gather online behavioral data and store it for a set period of time (e.g. one year) in accordance with GDPR regulations.

“second, it allows them to gather online behavioral data and store it for a set period of time (e.g. one year) in accordance with GDPR regulations (Trocin et al., 2021a).”

• cites Trocin et al., 2021aNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

Humans will be able to evaluate results and detect any outliers or anomalies in processed data.

“In other words, humans will be able to evaluate results and detect any outliers or anomalies in processed data.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Emotional intelligence is not part of these AI systems.

“It is worth mention that emotional intelligence is not part of these systems although understanding how people deal with emotional challenges is crucial for AI systems to emulate human reasoning (Luong et al., 2021).”

• cites Luong et al., 2021 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

The proposed model suggests that although there are inhibitors and barriers and despite the different ways of approaching AI governance, it offers positive outcomes if best practices are followed.

“Our proposed model suggests that although there are inhibitors and barriers and despite the different ways of approaching AI governance, it offers positive outcomes, if best practices are followed, and this study identified specific procedural, structural and relational components that are necessary for achieving this.”

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Managers having extended conversations with employees of other departments ensuring them that AI will not replace them could be crucial for the company's internal stability, as people might lose trust in the leadership, leave the company taking their expertise with them, or resist using new technologies and try to undermine the value of AI.

“This could be crucial for the company’s internal stability as people might lose trust in the leadership, they might leave the company taking their expertise with them or resist using new technologies and try to undermine the value of AI.”

May combine two claims.

The study used a multi-case study of three organizations that have been using AI for several years, with a series of interviews with key respondents.

“Through a multi-case study of three organizations that have been using AI for several years, we conducted a series of interviews with key respondents and identified a set of activities that were relevant under each of the three dimensions”