Figure 6 - uploaded by David Autor
Content may be subject to copyright.
More than 60% of jobs done in the United States in 2018 had not yet been "invented" in 1940 Source: Autor et al. (2022).
Context in source publication
Citations
... Every facet of the operation must be assessed for vulnerabilities. Individuals and organizations frequently make adjustments with little knowledge or understanding of the effects of the adjustments they initiate or endorse [11,12]. ...
In the rapidly evolving landscape of education, creative problem-solving (CPS) has become a vital competency for effective school management. This paper examines how educators and administrators employ CPS strategies to address multifaceted challenges within schools, from individualized student learning needs to systemic organizational reforms. By synthesizing theoretical models, empirical research, and real-world case studies, the study highlights the essential role of creativity in leadership, collaborative decision-making, and operational efficiency. It further examines the influence of technology, resistance to change, and the importance of fostering a creative mindset among staff. Through practical frameworks and evidence-based techniques, this paper provides actionable insights into implementing sustainable changes that enhance educational outcomes. Ultimately, it positions CPS not only as a tool for navigating complexity but as a catalyst for innovation and cultural transformation within educational systems. INTRODUCTION Here and now is a crucial time for educators. A teacher must focus on their students, understanding their needs, misunderstandings, and how to reach everyone effectively. To tackle these challenges, educators engage in a creative problem-solving process, managing information, their environment, and the students themselves. Enhancing creative problem-solving skills enables teachers to be better managers of student learning. Educators enter the profession for various reasons, such as a love for children, a desire to create meaning, and to facilitate learning. Learning is complex and not yet fully understood by research, which adds to the nobility of teaching. Today, high-stakes materials and tests have diminished the philosophical aspects of teaching, often reducing it to merely good classroom management. However, effective management encompasses creative problem solving, addressing individual student needs for varying approaches to learning. It involves visual awareness of classroom activities and auditory perception of student interactions, along with attention to environmental factors affecting students. Good management includes being present among students, offering support, adjusting the classroom dynamics, and maintaining respect during students' nervous moments while demonstrating confidence and enthusiasm in the learning process. School administrators now undergo rigorous selection processes for content knowledge and management skills. Research has shown that educators can be taught creative problem-solving techniques effectively. Yet, a meta-analysis found that only one study assessed the training's effectiveness specifically for school administrators. Among 16 studies on the professional development of educators in creative problem solving, officials reported strong content knowledge backgrounds [1, 2]. Importance of Creative Problem-Solving in Education A creative problem-solving approach in education allows stakeholders to collaboratively address longstanding challenges. Given the complex environments of modern education, a broader perspective is essential. The emphasis on quality, knowledge, and culture in today's intelligent organizations necessitates new frameworks for thought and action. The creative potential in school management is significant, as illustrated by in-depth case studies that align with key research questions regarding change processes. The urgency for educational improvement in a globalized context, where knowledge and
... There is a variety of explanations for and discussions of how AI can influence the labour market, also implying that it can be difficult to be explicitly clear about the impact carried by AI in this way (Autor, 2022). ...
... Nevertheless, it also presents a wide range of challenges for welfare states in the coming years, including loss of jobs and biases in the provision of welfare services. The possibility of specifying the impact of AI is limited, although there is potential to improve job quality, as well as assisting with climate change, diseases etc. (Autor, 2022). ...
... Lastly, the earnings effect of occupational licensing may vary across workers or occupations with different levels of skills. Our study focuses on the potential of dynamic, heterogeneous changes in the earnings effect of occupational licensing, given growing demand for skilled workers and healthcare workers in licensed occupations over recent decades (Autor, 2022;Catlin and Cowan, 2015). ...
... This is known as the Polanyi's paradox. The job polarization evidence led Autor (2022) to imagine a new division of labor, where machines and algorithms replace humans in any conceivable task that can be narrowed down to a set of instructions or procedures. The issue is that such codifiable activities are expanding fast, due to the impressive capabilities of AI. ...
This paper contributes to the recent literature on the economic growth implications of automation and artificial intelligence (AI), by devising and exploring the dynamics of an analytical model featuring the gradual substitution of labor by automated processes. In the proposed setting, a symbiotic relation between the expansion of AI and the accumulation of human capital stands out as the driver of long-term sustained growth. Two versions of the model are scrutinized. In the first version, the representative capitalist is an optimal planner, while workers (who are heterogeneous regarding their productivity levels) are hand-to-mouth consumers. In the second version, every agent formulates an intertemporal plan and may, simultaneously, be a worker and an investor. This difference in context is vital for identifying the potential beneficiaries of the implementation of labor-saving technologies.
... However, this focus has also led to some negative consequences regarding human considerations and impact on employees. Standardisation of processes through DAI has been linked with lower levels of work engagement (Gerdes, 2008), while automation of processes is impacting rates of employment (Autor, 2022) and employee identity (Moulaï et al., 2022). Additionally, diminished social connections at work triggered by this technology have negative implications for employee satisfaction (Gonzales, 2023). ...
Generative AI (GenAI), a subcategory of Artificial Intelligence (AI), leverages machine and deep learning algorithms to produce original content in various formats, including text, images, video, and sound (e.g., ChatGPT, Midjourney). An intuitive interaction through natural language commands (prompts), in combination with an element of surprise in the generated output, has fuelled widespread enthusiasm for GenAI. Yet the quality, originality, and value of the output vary based on user commands and training data. This often raises questions about consistency and optimal use of GenAI, especially compared to the established uses of AI. Meanwhile, while the predominant application of AI in the workplace is focussed on predictable algorithmic models for task automation and work efficiency, this attention towards automation has also led to questions about meaningful work and employee contribution at work. In this chapter, we explore the emerging affordances and malleability of GenAI to produce original and useful output, often associated with a creative act. To differentiate it from the tool-for-the-task usage of AI (e.g., credit rating), we position GenAI as a medium—a channel for exploration, innovation, and creative experience at work—a shift to more human-oriented work participation in a digitalised environment.
... 6 shows Zeta's implementation impact across different crop types and growing regions, highlighting how AI effectiveness varied by context. ...
This publication examines the perplexing disconnect between rapid advancements in generative AI technologies and the absence of corresponding gains in macroeconomic productivity measures phenomenon we term the "Modern AI Productivity Paradox." Through a comprehensive analysis spanning historical precedents, measurement challenges, implementation barriers, and distributional effects, we develop a multi-faceted framework for understanding this paradox. Our research combines quantitative economic analysis with qualitative case studies to demonstrate how the full economic impact of transformative AI technologies follows complex diffusion patterns that current metrics fail to capture adequately. We identify critical complementary investments and institutional adaptations necessary for AI's productivity potential to materialize at scale. The findings suggest that AI's economic impact will likely follow an S-curve adoption pattern similar to previous general-purpose technologies, but with potentially steeper acceleration once implementation barriers are overcome. This work contributes to both economic theory on technological transitions and practical policy approaches to harness AI for broadly shared economic prosperity in the coming decade.
... In her study on Western Germany, Spitz-Oener (2006) identifies a similar trend for the years 1979-1999, which can be shown to persist at least until 2018. 1 Observed changes in task composition are closely linked to the adoption of new technologies, a relationship conceptualized in the taskpolarisation literature. This literature acknowledges that technological progress can replace labourthough not all types to the same extentwhile complementing some tasks and substituting others (Autor 2022). Routine manual, cognitive, and service tasks are more likely to be replaced by new technologies, whereas non-routine manual, cognitive, and interpersonal activities are more likely to be enhanced (Acemoglu and Autor 2011, Autor and Dorn 2013, Goos, Manning, and Salomons 2014. ...
... To that end, we use data from the German pension insurance, covering the entire population of 2.4 million insured individuals from birth cohorts around the onset of the reform. The data includes detailed monthly employment status information for individuals aged 2 Recent developments in artificial intelligence, which may enable the substitution of non-routine tasks, as discussed in Autor (2022), have not been relevant for the birth cohorts studied here. 3 Leaving the workforce through a favourable retirement option might be particularly attractive for older employees with a high share of routine tasks, as they tend to transition into lower-skill jobs when their skills become outdated (Autor and Dorn 2009). ...
We study heterogeneity in the effects of a pension reform in Germany: the introduction of the old-age pension for the very long-term insured, which lowered the threshold for full pension receipt from age 65 to 63. Using a regression discontinuity design (RDD) and rich administrative pension data, we estimate the effects of the reform on eligible workers who differ in a number of occupational characteristics. Transitions into the new early retirement scheme occurred more frequently from jobs involving mainly manual and routine tasks, which contributed to the changing composition of tasks in the workforce. While the introduction of new technologies, materials and machines in an industry was associated with more workers claiming early retirement, workers affected by frequent PC use and new PC programmes were less likely to use the early retirement scheme.
... Second, there is work on cooperation in networks with repeated interactions (usually playing a Prisoner's Dilemma with neighbours). Early contributions here typically consider patient players (Kandori, 1992;Ellison, 1994), with later contributions considering impatient agents (Ali and Miller, 2013, 2022. 3 My setting differs from these strands in two important ways. ...
... It does not rely on strategic inter-dependencies in neighbours' actions (as in the public goods strand) or on repeated interactions 2 A review of the large literature on the harmful effects of innovation is beyond the scope of this paper. Some recent contributions here include Autor (2022); Acemoglu (2023); Johnson and Acemoglu (2023). 3 Other important contributions to this literature include Vega-Redondo (2006); Lippert and Spagnolo (2011); Nava and Piccione (2014) and Jackson et al. (2012). ...
This paper presents a model of network formation and public goods provision in local communities. Here, networks can sustain public good provision by spreading information about people's behaviour. I find a critical threshold in network connectedness at which public good provision drops sharply, even though agents are highly heterogeneous. Technology change can tear a community's social fabric by pushing high-skilled workers to withdraw from their local community. This can help explain rising resentment toward perceived ``elites'' -- their withdrawal actively harms those left behind. Moreover, well-meaning policies that upskill workers can make them worse off by reducing network connectedness.
... Conventional econometric models, as functional as they are, tend to fall short when grasping the sophistication and fluidity of economic conditions. Machine learning (ML) methods offer a substantial alternative: with the help of extensive data and the discovery of sophisticated patterns that go unperceived by standard models [1]. ...
The COVID-19 pandemic resulted in one of the most recent economic shocks, impacting global trade, financial markets, and consumer behavior. In the US, GDP suffered a historic downturn in 2020, followed by an unbalanced recovery. Precise GDP growth forecasting has become increasingly essential for policymakers, businesses, and investors making decisions in the post-pandemic economy. Classic models, including Autoregressive Integrated Moving Average (ARIMA), Vector Autoregression (VAR), and Dynamic Stochastic General Equilibrium (DSGE), have been popularly employed for GDP forecasting. Machine learning (ML) provides a dominant alternative, with the potential to handle enormous amounts of real-time data, sense non-linear patterns, and handle economic shocks more effectively than traditional approaches. This paper delves into the potential of ML in GDP forecasting, touching on some key techniques, including neural networks, ensemble learning, and deep learning. This paper assessed the accuracy of two machine learning models, Random Forest (RF) and Long Short-Term Memory (LSTM), in forecasting U.S. GDP growth during the post-COVID-19 pandemic. Although ML-based forecasting holds prominent advantages, challenges, including data quality, explainability, and ethical issues, must be resolved for increased usage in economic decision-making.
... Conventional econometric models, as functional as they are, tend to fall short when grasping the sophistication and fluidity of economic conditions. Machine learning (ML) methods offer a substantial alternative: with the help of extensive data and the discovery of sophisticated patterns that go unperceived by standard models [1]. ...
The COVID-19 pandemic resulted in one of the most recent economic shocks, impacting global trade, financial markets, and consumer behavior. In the US, GDP suffered a historic downturn in 2020, followed by an unbalanced recovery. Precise GDP growth forecasting has become increasingly essential for policymakers, businesses, and investors making decisions in the post-pandemic economy. Classic models, including Autoregressive Integrated Moving Average (ARIMA), Vector Autoregression (VAR), and Dynamic Stochastic General Equilibrium (DSGE), have been popularly employed for GDP forecasting. Machine learning (ML) provides a dominant alternative, with the potential to handle enormous amounts of real-time data, sense non-linear patterns, and handle economic shocks more effectively than traditional approaches. This paper delves into the potential of ML in GDP forecasting, touching on some key techniques, including neural networks, ensemble learning, and deep learning. This paper assessed the accuracy of two machine learning models, Random Forest (RF) and Long Short-Term Memory (LSTM), in forecasting U.S. GDP growth during the post-COVID-19 pandemic. Although ML-based forecasting holds prominent advantages, challenges, including data quality, explainability, and ethical issues, must be resolved for increased usage in economic decision-making.
