Periodic Reporting for period 1 - AIFocus (The causes and consequences of firms’ focus on artificial intelligence)
Reporting period: 2023-01-01 to 2024-12-31
Artificial intelligence (AI) has become a pervasive phenomenon with a profound impact on economies, societies, and organizations. Examples include natural language processing, advanced machine learning, and robotic process automation. AI is estimated to contribute to the global economy through productivity gains. More importantly, AI has the potential to solve some of the world’s greatest challenges through innovation.
Two core AI capabilities are AI automation and AI information. Firms use AI to increase efficiency through automating their business processes (AI automation). Software tools such as robotic process automation allow firms to automate workers’ repetitive tasks. These technologies rapidly speed up work processes and drive down costs. Firms also try to foster innovation by discovering new information (AI information).
While the potential of AI is generally acknowledged, it is not evident that the focus on AI will translate into the targeted economic benefits. There is limited empirical evidence about the firm-level causes and consequences of firms’ increasing focus on AI. The reason for this lack of knowledge is a paucity of firm-level data on AI focus.
This project contributes to our understanding of the causes and consequences of AI by collecting and analyzing unique firm-level data. Understanding the causes and consequences of firms’ AI focus is important because it informs policymakers about the role of AI in the economy and can help policymaking concerning AI.
Two core AI capabilities are AI automation and AI information. Firms use AI to increase efficiency through automating their business processes (AI automation). Software tools such as robotic process automation allow firms to automate workers’ repetitive tasks. These technologies rapidly speed up work processes and drive down costs. Firms also try to foster innovation by discovering new information (AI information).
While the potential of AI is generally acknowledged, it is not evident that the focus on AI will translate into the targeted economic benefits. There is limited empirical evidence about the firm-level causes and consequences of firms’ increasing focus on AI. The reason for this lack of knowledge is a paucity of firm-level data on AI focus.
This project contributes to our understanding of the causes and consequences of AI by collecting and analyzing unique firm-level data. Understanding the causes and consequences of firms’ AI focus is important because it informs policymakers about the role of AI in the economy and can help policymaking concerning AI.
The first part of the project focused on developing a firm-level measure that captures firms’ focus on AI. To achieve this goal, the project measures firms’ focus on AI across publicly listed firms by analyzing managers’ communication with their stakeholders in earnings conference calls. In this communication, firms regularly discuss their current business operations, strategic plans, and investments.
I created three AI measures using different, complementary methods. The first AI measure of the project was developed based on textual analysis using a dictionary of AI-related terms. Next, I create two more advanced AI measures using large language models (LLMs). I fine-tuned a Bidirectional Encoder Representations from Transformers (BERT) model using 6000 human annotated sentences. Then, I use this fine-tuned model to classify AI sentences in firm communications, resulting in a second AI measure. The third measure was developed based on a new powerful textual analysis method using an open-source generative LLM. I also expanded the measures to capture additional information about firms’ AI focus (e.g. AI for automation, AI for information, AI investment). The three measures built the foundation for the remaining project and are an important outcome.
The second part of the project aimed to investigate the causes and consequences of firms’ AI focus. Specifically, I investigate the effect of unionization on firms’ AI focus. Labor unions can substantially impact firms’ operating flexibility and strategic decisions. Yet, their impact on firms’ AI initiatives has received scarce attention in the literature. To examine how labor unions influence firms’ AI focus, I collect and construct five measures to capture firm-level unionization. Another innovation of the project is the construction of firm-level unionization based on firms’ annual reports using an LLM comparable to the approach used for the AI focus measure. After collecting financial and governance data, I conduct extensive econometric analysis. The results suggest that unionized firms have lower levels of AI focus than their non-unionized counterparts. The working paper is currently being prepared and will be publicly available in 2025. Additionally, the project focuses on the firm benefits of AI in analyzing data and integrating knowledge. In this part, I exploit the AI measures developed in the first part of the project. A working paper on the final empirical part of this project will be completed at the end of 2025.
I created three AI measures using different, complementary methods. The first AI measure of the project was developed based on textual analysis using a dictionary of AI-related terms. Next, I create two more advanced AI measures using large language models (LLMs). I fine-tuned a Bidirectional Encoder Representations from Transformers (BERT) model using 6000 human annotated sentences. Then, I use this fine-tuned model to classify AI sentences in firm communications, resulting in a second AI measure. The third measure was developed based on a new powerful textual analysis method using an open-source generative LLM. I also expanded the measures to capture additional information about firms’ AI focus (e.g. AI for automation, AI for information, AI investment). The three measures built the foundation for the remaining project and are an important outcome.
The second part of the project aimed to investigate the causes and consequences of firms’ AI focus. Specifically, I investigate the effect of unionization on firms’ AI focus. Labor unions can substantially impact firms’ operating flexibility and strategic decisions. Yet, their impact on firms’ AI initiatives has received scarce attention in the literature. To examine how labor unions influence firms’ AI focus, I collect and construct five measures to capture firm-level unionization. Another innovation of the project is the construction of firm-level unionization based on firms’ annual reports using an LLM comparable to the approach used for the AI focus measure. After collecting financial and governance data, I conduct extensive econometric analysis. The results suggest that unionized firms have lower levels of AI focus than their non-unionized counterparts. The working paper is currently being prepared and will be publicly available in 2025. Additionally, the project focuses on the firm benefits of AI in analyzing data and integrating knowledge. In this part, I exploit the AI measures developed in the first part of the project. A working paper on the final empirical part of this project will be completed at the end of 2025.
The research conducted in this project has advanced beyond the state of the art in two key areas. First, the project developed a novel measure of firms’ AI focus based on public communication. Existing research has predominantly focused on national-level and industry-level data or relied on survey evidence. The AI focus measure captures AI focus at the firm level and allows other researchers to track AI uptake and investigate new questions related to firms’ AI focus. The measure can be annually updated. Second, the project advances our understanding of the factors that impact firms’ AI focus (i.e. unionization) and firm-level benefits (e.g. integrating knowledge for innovation).