Written by GCS research analysts Eva Yang and Leo Mak.

Introduction

Recent years have seen rapid advancements in technology, and its impact on the labour market has been significant. By now, you have probably heard of Chatgpt, the AI chatbot prototype capable of generating fluent responses to human inputs. Its potential applications are wide-ranging, from writing essays to composing rap songs. In fact, tech giants like Google and Bing are actively trying to implement similar language models into their search engines to provide a more natural & intuitive search experience for their users. While these developments in natural language processing have the potential to revolutionise various industries, there are also concerns about their impact on employment and other societal issues. Therefore, it is important to carefully examine the potential benefits and drawbacks of ChatGPT and similar technologies. Therefore, the following questions will be explored in this paper: How does the use of ChatGPT and similar technologies affect employment and job displacement? What are the potential benefits and drawbacks of these technologies from an economic perspective? By carefully examining the potential impact of ChatGPT and similar technologies, we can better understand how to maximise their benefits while minimising any potential negative consequences. To explore these questions, this paper will be structured as follows. First, we provide a definition of ChatGPT and an explanation of how it works. We then review the existing literature on the impact of automation on employment, as well as the impact of NLP on employment. We also discuss the relevant economic theories that help explain the relationship between technology and employment. Next, we present our findings and interpret them in the context of previous research. Finally, we discuss the policy implications of our results.

NLP Evolution and ChatGPT: Exploring the Future of Language Technology

When asking Chatgpt about the definition of what Chatgpt is and its response is: ” ChatGPT is a type of NLP technology that uses deep learning algorithms to generate human-like responses to text inputs. It was developed by OpenAI and is based on a transformer-based neural network architecture. ChatGPT is trained on large amounts of text data and can generate responses to text inputs in real-time.”

The backbone technology that supports Chatgpt is Natural Language Processing. It is a field of study that combines computer science and linguistics to enable machines to understand, interpret, and generate human language. The development of NLP dates back to the 1950s, with the creation of early language translation programs. In the 1970s, the development of syntactic and semantic parsing techniques paved the way for more advanced NLP applications. Since then, NLP has found widespread use in a variety of industries, including healthcare, finance, customer service, and education. In healthcare, NLP is used to analyse patient data and support clinical decision-making. In finance, NLP is used to analyse news and social media to inform trading decisions. In customer service, NLP is used to automate responses to customer inquiries. In education, NLP is used to develop intelligent tutoring systems that can adapt to individual student needs.

Literature review

Over the past few decades, there has been a growing body of research examining the impact of automation on employment. Cardullo and Ansal (2016) argue that automation replaces routine jobs and increases efficiency and productivity, leading to job loss. However, they suggest that new jobs that require high levels of skills, such as technical skills, can be created. Frey and Osborne’s (2017) study found that approximately 47% of jobs in the US are at high risk of being automated. They argue that new job opportunities will require skills that cannot be easily automated, such as creativity and complex problem-solving. Similarly,one study by Piero Formica (2018) investigates the future of employment in the AI era and argues that new skills will be required for workers to remain employable. The study highlights the importance of creativity, emotional intelligence, and social skills as key competencies for workers in the future labour market

The impact of NLP on employment has also been studied by researchers, with some suggesting that it can lead to job polarization and skill-biased technological change. For instance, in a study by Autor, Levy, and Murnane (2003), the authors found that computerization and automation led to a polarization of the labour market. They argue that these technologies can perform routine tasks more efficiently than humans, leading to a decline in the demand for middle-skill jobs while increasing the demand for low-skill and high-skill jobs. This trend is further explored by Daron Acemoglu and his colleagues. They analyzed online job vacancies in the United States from 2010 onwards and discovered that establishments exposed to AI are selectively removing vacancy postings that mention a range of skills that were previously required while adding new skill requirements that were not listed before. However, the study also indicates that AI is currently replacing human workers in only a limited range of tasks, and the overall impact on the labour market is not yet evident.

Discussion

Traditional automation

The literature reviewed above highlights the potential impact of automation and AI on employment and the labour market. The implication of the studies is that the increasing use of these technologies has the potential to lead to job polarization and skill-biased technological change. This is a significant concern as it may lead to a decline in middle-skill jobs, which can exacerbate income inequality and reduce social mobility.

These changes in the labour market can be explained by the Solow Growth Model. The model suggests that technological progress can lead to an increase in labour productivity, which can lead to economic growth. However, this growth can also lead to a shift in the demand for labour, where high-skill workers are in greater demand, while middle-skill workers are less in demand. This can result in wage inequality and a decline in middle-skill jobs. These changes in the labour market can be exacerbated by the use of automation and AI, which can perform routine tasks more efficiently than humans, leading to a decline in demand for middle-skill jobs.

The Structural Change model can also provide insights into the potential impact of automation and AI on the labour market. The Structural Change model suggests that changes in technology and the economy can lead to shifts in the composition of the labour force. As industries evolve, some sectors may decline while others grow, leading to changes in the demand for labour in different industries. This can lead to challenges for workers who may need to acquire new skills to remain employable.

With the increasing adoption of automation and AI, certain industries, forecasted by PWC China, may experience rapid growth, while others may decline. For example, automation and AI could lead to the decline of certain industries, such as manufacturing (projected -36% job displacement), while other industries, such as healthcare and technology, may experience growth. As a result, workers may need to adapt to changes in the labour market, which could involve acquiring new skills and transitioning to new industries.

Comparing traditional automation to Chat GPT in the Chinese Labour Market

For instance, when we look at the automatic supply chain system, we can construct an ETL/SQL to do the data cleaning task for the employees. As long as the initial framework has already been built, it can be used once and for all. This means, there is no longer a need for employees in the data sorting process anymore.

When it comes to Chat GPT, it can rarely replace the physical labour force but can improve the work frame of the white collar worker who is doing dull and repetitive paperwork. For instance, it can inspire the media PR with detailed but already existing planning proposals in a limited time. However, it can only make the boring work more efficient, and cannot update the database with new and creative ideas on its own. Put it in a simple way, you can tell ChatGPT to create a van gogh style painting, but there must be a van gogh first to draw his own picture and create the style that is unique to him. Therefore, it can hardly replace the labour force at this technology level. 

Recommended policies for China

To address these issues, policymakers need to develop strategies that mitigate the potential negative impact of AI and automation on employment. One approach is to invest in education and training programs that provide workers with the skills and knowledge necessary to compete in the job market of the future. This can include programs to help workers transition from low-skill jobs to high-skill jobs, as well as programs that provide workers with specialized technical skills needed to work with AI and other advanced technologies.

Another approach is to promote policies that incentivize the development and adoption of AI technology in a way that benefits workers and the broader economy. This can include tax credits and other incentives for businesses that invest in training and re-skilling programs for workers, as well as policies that support the creation of new industries and job opportunities.

In addition, policymakers should consider implementing social safety net programs to provide support to workers who are displaced by automation. This can include unemployment insurance, job search assistance, and income support programs.

Conclusion

The article discusses the rapid advancements in technology, specifically the ChatGPT AI chatbot prototype and its impact on the labour market. The potential applications of ChatGPT are wide-ranging, but there are concerns about its impact on employment and other societal issues. The article examines the potential benefits and drawbacks of ChatGPT and similar technologies from an economic perspective. Meanwhile, with the right policies and investments, it is possible to mitigate the negative impact and create new opportunities for workers and the broader economy.

Reference

Autor, David H., Frank Levy, and Richard J. Murnane. “The Skill Content of Recent Technological Change: An Empirical Exploration.” The Quarterly Journal of Economics 118, no. 4 (2003): 1279–1333. http://www.jstor.org/stable/25053940.

Acemoglu, Daron, David Autor, Jonathon Hazell, and Pascual Restrepo. “AI and Jobs: Evidence from Online Vacancies.” National Bureau of Economic Research, working paper no. 28257, December 2020, doi: 10.3386/w28257, URL: http://www.nber.org/papers/w28257.

Cardullo, F., & Ansal, A. (2016). A Survey of Natural Language Processing Techniques. Journal of Artificial Intelligence and Soft Computing Research, 6(1), 1-11.

Formica, P. (2021). Human manufacturing and the challenges of AI. Industry and Higher Education, 35(3), 147–149. https://doi.org/10.1177/09504222211000279

Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation?. Technological Forecasting and Social Change, 114, 254-280.

Hawksworth, John, Yuval Fertig, Stephan Hobler, Richard Berriman, Hugh Dance, Anand Rao, Massimo Pellegrino, Daniel DiFilippo, and Jonathan Gillham. “What Will Be the Net Impact of AI and Related Technologies on Jobs in China?” PWC, September 2018. https://www.pwc.com/gx/en/issues/artificial-intelligence/impact-of-ai-on-jobs-in-china.pdf. 

Soares, A., & Yacef, K. (2019). A review of natural language processing in education: Opportunities and challenges. IEEE Transactions on Learning Technologies, 12(4), 455-468.

王志凯. “人工智能等技术进步对就业的影响——基于三次产业的实证分析.” 时代金融 33(2018):3.