ChatGPT: a pleasant surprise and a distress signal
“Will my job be replaced by ChatGPT?” This question might haunt millions of users when they are testing how well the programme could work.
ChatGPT has gone viral on the Internet. This OpenAI-developed chatbot is able to generate responses in a human-like conversational format. Simply by putting your questions or requests in the box and clicking, it is capable of providing solutions to a wide range of topics from composing sonnets to developing preliminary business strategies, writing codes to drafting legal documents within seconds. Since its launch in late November 2022, it has gained over 100 million monthly users[i]. This meteoric rise was unprecedented in the history of any internet app.
(Source: Getty Images)
Unlike many other automated services that only handle routine and repetitive work, ChatGPT is capable of doing more sophisticated and creative jobs. Analysing data, processing information, commenting codes, writing emails, iterative learning, and generating fair essays. It never dodges difficult questions and even passed an undergraduate labour economics mid-term test recently given by Prof. Bryan Caplan from George Mason University[ii]. It also passed the Evidence and Torts section of the National Conference of Bar Examiners (NCBE) Multistate Bar Examination (MBE) practice exam and is presumed to pass the MBE soon[iii]. The productivity it yields, and the structured, well-informed content it produces have never ceased to amaze the audience.
Meanwhile, it has stirred a huge wave of shock, especially among knowledge workers. According to the DIKW pyramid[iv], knowledge is considered as “high value”, where skilled workers, often white collars, produce services or products such as codes, marketing materials, customer service, etc. to spark innovation and growth in the company. They possess diverse skill sets, differentiate products and services, and add economic value to the organization. The increased productivity and effectiveness of knowledge workers is the engine for economic growth[v].
The release of ChatGPT has helped white collars improve their performance: businesses spend less time creating marketing materials, and lawyers can produce legal briefs much faster. For employers, the tremendous amount of information and content it produces in an instant and cheap manner allows them to make increasingly marginal profits. For employees, despite ChatGPT can improve their productivity at present, the risk of certain skills being undervalued is high and chances of employment would be hurt in the long run. As ChatGPT’s capability of producing content is expected to polish over time, an obvious challenge is presented: how far will it go and how long will it take to wipe out jobs?
The destabilization of the employment landscape and widening inequality
ChatGPT is not the only player in the game of AI technologies. Google-developed Bard AI will be made to the public in the coming weeks[vi] as this article is written. Alibaba, a Chinese Internet giant, is undergoing internal testing to rival ChatGPT[vii]. Advancements in artificial intelligence technologies have long captured the imagination of many since the 1960s. How will it impact workers and the employment landscape? Will it wipe out jobs eventually?
(Source: Getty Images)
The impact of automation on employment has always been a heated debate. Technology optimists believe that the productivity effect outstrips the displacement effect[viii]. Once products or services are scaled to an extent that is affordable to as many consumers as possible, firms aim for more workers and capital. Firms are also incentivized to expand their businesses, thus recruiting more skilled employees to perform non-automated jobs. The demand for workers with the knowledge to develop and manage systems is growing. It can create more jobs for healthcare workers as life expectancy becomes higher and the population is aging. As the pie of the national economy grows bigger, a smaller share can greatly benefit people. The job losses in particular sectors can be offset by the broadly-based prosperity.
However, the interplay of socioeconomic factors and technologies only complicates the employment landscape. As automated technologies rely on machines to perform a variety of tasks by humans, the value created by laborers is decreased, thus declining employment opportunities and wages. Compared with highly skilled experts, undereducated workers, and marginalized groups are more vulnerable to the impacts of automation. They are less likely to migrate to other places and find jobs that match their skills. The situation is particularly pronounced in a developing country, where undereducation is more prevalent. According to ILO [ix], about 70% of employees with a low level of educational attainment or insufficient qualifications fail to meet the job requirements. More women are undereducated and are likely to secure jobs that match their education. As a result, the gap between the global north and global south countries is expected to widen, and inequality within the country is growing.
Technologies in AI have transformed jobs and societies in a way that no one has predicted. AI makes service sectors such as construction, health care, and business more efficient, where routine and repetitive work are substituted. Much as ChatGPT can empower individuals, it can present a threat to the existing knowledge economy. In developing countries, education has played an important role in promoting human capital and social mobility. Yet ILO suggests only a quarter of aspiring graduands can find a job that matches their education level in low-income countries. The shrinking job opportunities as knowledge workers further lead to what Chong[x] calls the “devaluation of valuable labor” and further leads to downward mobility like factory workers.
Digital inclusion remains a challenge. According to World Bank, around three billion people are offline, and the majority of them are in developing countries[xi]. Under the context of a huge gap in the digital divide and AI divide, a new wave of billionaires will gain massive wealth while many white-collar workers will be pushed out of well-paid jobs. The widespread job loss, the reduction of working hours, and stagnant wages in less privileged groups and developing countries are likely to exacerbate the existing inequality. The application of AI itself is a manifestation of raising productivity, whereas the trickling down to the individual level can be painful. The disproportionate economic and political dominance of the business world can lead to popular discontent and lead civil movements.
How will the world change by 2050? It is difficult to make a precise prediction about how the maturation of AI like ChatGPT will impact the workforce, as different industries and jobs will require different sets of skills, but it is concerning that before the increase in productivity increases can benefit society and improve our lives, humans are likely to earn less than expected or even be unemployed in the future. As Paul Krugman[xii] suggests, the impact on overall employment can be hard to identify.
Call for policy interventions
ChatGPT can unleash the potential of knowledge workers: by freeing up them from tedious and repetitive work and allowing space for inspiration and innovation. In the meantime, it can bring unprecedented effects on the workforce and society.
Technological advances are constantly reshaping work and the demand for skills. In the neoliberal era, individuals are expected to take full responsibility for themselves and prepare for the future as the wheel of history is inexorably turning. Interpersonal and emotional skills are required more than ever, and both cross-sectoral and vertical skills will drive their careers and the industry forward. Individuals can develop their insights and plan careers based on their judgments. In the meantime, the government ought to invest in human capital, providing upskilling and reskilling opportunities to ensure workers have the right skills, as well as a comprehensive dataset that allows jobseekers to have more job information as well as match jobseekers’ skills and abilities to the right jobs.
The technological innovations represented by ChatGPT will not only impact the workforce and generate inequality, but also provokes societal changes at an unknown scale. It is critical that policymakers and business leaders carefully consider its impacts on the economy and the labor market, expand social protection and adopt the appropriate regulatory framework.
References
[ii] https://betonit.substack.com/p/chatgpt-takes-my-midterm-and-gets
[iii] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4314839
[iv] https://www.wikiwand.com/en/DIKW_pyramid
[v] Davenport, T. H. (2005). Thinking for a living: how to get better performances and results from knowledge workers. Harvard Business Press.
[vi] https://blog.google/technology/ai/bard-google-ai-search-updates/
[viii] Acemoglu, D., & Restrepo, P. (2020). The wrong kind of AI? Artificial intelligence and the future of labor demand. Cambridge Journal of Regions, Economy, and Society, 13(1), 25-35.
[x] Chong, K. (2018). Best practice: management consulting and the ethics of financialization in China. Duke University Press.
[xi] https://www.worldbank.org/en/topic/digitaldevelopment/overview
[xii] https://www.nytimes.com/2022/12/06/opinion/chatgpt-ai-skilled-jobs-automation.html
