Artificial intelligence will replace some jobs and create new ones

The explosion of artificial intelligence（AI） software such as Chatgpt 4 and Midjourney has brought attention back to the field of AI, and their ability has caused many people to feel the crisis, and it is. This January, Microsoft Corporation announced that it will lay off 10,000 employees, firing the entire ethics and society team in its AI division. (Novet, 2023) Hopes to cut costs amid economic uncertainty and refocus on strategic priorities such as AI. At the same time, a host of gaming companies have been quick to embrace technological change and introduce AI painting into their workflows to escape the huge talent pressures and funding anxieties of the gaming industry. This article argues that as society changes, it is inevitable that AI will become more involved in jobs and replace some people and positions. However, recent events related to a naked photo of a woman on a subway going viral (China youth news, 2023) and nearly 400 crashes in the US in 10 months involving cars using advanced driver assistance technology have brought renewed attention to the potential risks of AI. Therefore, there are sufficient reasons to believe that current AI technologies can indeed replace some work but not all. Firstly, I will briefly introduce what AI is and how it works. Secondly, I will analyze the potential risks of current AI technologies. Finally, a discussion of the careers that may be replaced by AI and future career trends will be developed around the risks of AI.

What is AI？

Artificial Intelligence (AI) refers to the creation and use of algorithms to build dynamic computing environments that simulate the foundations of human intelligence processes. The systems of AI use mathematics and logic to simulate the reasoning that people use to learn new information and make decisions. In brief, the ultimate goal of AI is to allow computers to think and act like humans. Since the first century B.C., humans have been interested in the feasibility of building machines to simulate the human brain. In 1955, when John McCarthy coined the term ‘artificial intelligence’, people began to have a clearer name for this study. (Andresen, 2002) The following year, McCarthy and others organized a scientific conference that laid the foundations for AI. In modern times, algorithms and predictive analytics have emerged and given rise to several new careers and research areas. It has evolved to the current normative analysis. We see the application of AI in many everyday scenarios, like, financial services fraud detection. Banks use AI in the initial scoring of credit applications to understand the creditworthiness of applicants. There may not be just one answer to this question. What AI is “depends heavily on the goals of the researchers involved” (Schank, 1987), and any definition of AI depends on the methods used to build AI models.

How does AI work？

Virtual Customer Assistance (VCA) is widely used today. When a customer initiates a chat conversation on the web or over the phone, they usually start by interacting with a computer running specialized artificial intelligence. If the chatbot cannot explain or solve the problem, human customer service steps in and communicates directly with the customer. These unexplained instances are fed into a computing system that is used to improve the AI application to better complete the interaction. How are AI applications improved? To explain this one must mention machine learning（ML）, which is the process of using mathematical models of data to help computers learn without direct instructions. This allows computer systems to continue to learn and improve on their own based on experience. One way to train computers to mimic human reasoning is to use neural networks, a series of algorithms modeled on the human brain. (Axure, n.d.) The close connection between neural networks and computer systems is the basis for the implementation of artificial intelligence. It is essentially about artificial intelligence and machine learning working in tandem. Artificial intelligence uses machine learning to analyze and integrate information for eventual feedback to the user. For example, the recent explosion of ChatGPT is the result of OpenAI continuously giving ChatGPT data for machine learning, which ultimately allows it to recognize users’ questions and needs and give answers just like a human.

Potential risks of AI

The risks of AI are classified into the following three areas, accidental risks, structural risks, and misuse risks. (Zwetsloot & Dafoe, 2019) Given the complexity of structural risks, I will only discuss misuse risks and accidental risks here.

Misuse risk is the potential for people to use AI in unethical ways, and advances in AI technology have made it possible to “deepfake” not only photographs but even video or audio, which can be used for disinformation, political sabotage, or phishing attacks. The nude underground photo incident mentioned at the beginning of the article is good proof. With the explosion of ChatGPT, many students use ChatGPT to cheat on exams or assignments. A recent survey by Study.com shows more than a quarter of teachers found students using ChatGPT to cheat. (McDade, 2023) ChatGPT 4 has been able to score 90% higher on many exams, including the US Bar exam. (Dailymail, 2023) While teachers are concerned about students using AI to cheat, at the same time they believe that students using ChatGPT to ask questions can pique students’ curiosity and generate their questions. As a result, 2/3 of the people in the latest survey think that students should not be banned from using ChatGPT, especially if they cannot find the right answer in a Google search. (McDade, 2023) In my opinion, students will inevitably use AI in their studies in the future, as opposed to search engines and other learning methods that require more time to sift through the answers. Perhaps the next step for the education community is to think about how far the use of AI will be judged as cheating rather than banning it altogether.

Accident risk often involves injuries arising from AI systems behaving in unexpected ways. In 2016, a Tesla Model S failed to recognize a white truck on a clear day, leading to a collision between the two vehicles and causing the death of the Tesla driver. (Yadron & Tynan, 2016) Two years later, Uber’s self-driving car killed a pedestrian in Arizona, USA, after its machine learning system failed to account for jaywalking. (Griggs & Wakabayashi, 2018) At first, it was just thought that the system failed to detect the passerby. What has since come to light is quite shocking; the system itself was designed to see passers-by in plenty of time to stop the car before an accident occurred. Designed for safety, this made the system so sensitive that it could be triggered frequently to stop the vehicle, which would have made Uber’s operations less efficient, and the product would not have been competitive with other competitors when it was put on the market. As a result, the system was artificially turned off by the company’s engineers when it was finally put into use. The problems and risks of AI are so numerous and complex that human technology will not be able to break through to a perfect solution to these risks, at least not soon.

What are some of the jobs that are currently being replaced by AI?

The technology of AI is not yet mature, but it can already help companies and individuals do most of the tedious aspects of their work. As we can see from Tanya Tsui’s post, although it is a simple part and comes with a few bugs, ChatGPT does allow for independent coding guided by user requirements. (Tsui, 2023) In addition, Lost Lore, a Russian game studio, is currently working on a F2P mobile game called Bearverse. They used the AI software Midjourney for the character design phase, combining AI text input and manual adjustments to successfully reduce development costs from $50,000 to $10,000 and drastically reduce man-hours from six months to one month. (Semenov, 2023) This means that work that would have taken a team of people to complete, and which would have taken a lot of time and effort, could now be done with just one or a handful of people communicating effectively with the AI. What are some of the jobs that are currently being replaced by AI? 

–     Data analysts, where AI can process data faster, more comprehensively, and more accurately, whereas now data analysts need to collect and integrate and apply complex mathematical models to reach conclusions on their own. 

–     Primary content creators, such as in painting, film, and fiction, AI can use large databases to derive most human preferences to craft content that better suits their aesthetic, whereas at this stage it is difficult for content creators to create work that satisfies everyone because of individual differences. 

–     Teachers, who don’t think, don’t guide properly, and read exactly from a handout or PPT. 

–     Human resources, AI can do more fair and efficient recruitment, performance evaluation, and salary allocation to avoid subjective bias. 

–     Customer service, AI can quickly handle customer service with speed and efficiency. 

–     Marketing, AI can obtain and analyze market trends faster and develop marketing strategies, simplifying market research work. At this stage marketing departments still need to do more field market research work.

–     Chess players, for example, checkers, have more than 500 x 10 to the 18th power of the possible combinations. Very few humans can be considered masters, but artificial intelligence can calculate these combinations of moves with great efficiency and give the best possible response.

Will people lose their jobs because of AI?

Although the rapid development of AI and its powerful computing capabilities have caused many people to feel a crisis, I do not believe that AI will completely replace human work. Not only the three types of potential risks mentioned earlier but also humans have some complex emotions that machine learning cannot learn. 

I agree with Frank and Alina in mapping the future of occupation, which classifies occupations based on transformative effects and destructive effects, occupations are divided into “Rising stars” occupations, “Machine terrain” occupations, “Human terrain” occupations, and “Collapsing” occupations. (Fossen & Sorgner, 2019) The “Rising star” occupations, which require a high level of creativity and social intelligence, and the ” Human terrain” occupations, which are least affected by digitalization, cannot be replaced by machines. “AI is neither artificial nor intelligent.” (Crawford, 2021) AI is figurative and material, made up of concrete data and information, natural resources, human labor, etc. The learning mechanism of AI is such that it can connect different images, words, etc., re-edit them and put them together and feed them back to us rather than creating by thinking. This means that occupations such as judges, teachers, counselors, artists, and others that require a high level of empathy or creativity are destined to be irreplaceable. AI plays an atlas role in facilitating our lives, guiding us, and offering us new perspectives on re-understanding the world without “having to think that we are summarising or exhausting it.” (Crawford, 2021) 

Historically, since the development of technology, there has been the replacement of occupations, and the displacement of workers due to technological advances has been offset by the creation of new jobs. (Acemoglu & Restrepo, 2019) The development of AI has likewise given rise to several new occupations, for example, cue trainers, AI auditors, AI ethics experts, etc. Old jobs are also gradually transforming, with AI code organizers, AI graphic artists, and AI narrators corresponding to programmers, concept designers, and writers respectively in the past. In the interview, Li Xin mentions that without AI painting, I wouldn’t have come across this career because I don’t know how to paint, and I’m a programmer myself. (Chen, 2023) I find it very interesting that AI has lowered the barriers to entry to some professions, or perhaps lowered consumption. When a user wants an original avatar, he doesn’t need to know how to draw or spend a lot of money on an artist. He just needs to tap on the keyboard and the AI will give him a satisfactory answer. Overall, I am very happy with the development of AI, which helps us to achieve our goals more quickly and gives us a glimpse of the world’s many possibilities and a future full of variables.

Conclusion

Overall, the nature of AI is not to create but to integrate and many potential risks are currently unaddressed. We do not need to worry about AI replacing our jobs in the short term, but it is worth noting that humans and computers are generating huge amounts of data every day, far beyond the capacity of humans to absorb, interpret and make complex decisions based on it. This means that our mass use of AI is a necessary and inevitable future. Instead of worrying and being anxious that we will be replaced by AI, we should adapt as soon as possible and learn how to live with it. Let AI facilitate our lives and replace us with boring and repetitive tasks so that humans can have more energy to enjoy life and create new things.

Reference

Acemoglu, D., & Restrepo, P. (2019). Automation and new tasks: How technology displaces and reinstates labor (pp. 3–30). National Bureau of Economic Research. http://dx.doi.org/10.3386/w25684

Andresen, S. L. (2002). John McCarthy: father of AI. IEEE Intelligent Systems, 17(5), 84-85.

Axure. (n.d.). Artificial intelligence vs. machine learning. Microsoft Azure. Retrieved April 16, 2023, from https://azure.microsoft.com/en-us/resources/cloud-computing-dictionary/artificial-intelligence-vs-machine-learning/#introduction

Crawford. (2021). The Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press.

Chen Qiaohui. (2023, March 22). AI new career, what are they doing. Redian News. https://redian.news/wxnews/324011

Dailymail. (2023, March 27). Half of students are using ChatGPT to cheat, and it could rise to 90%. Daily Mail. https://www.dailymail.co.uk/sciencetech/article-11899475/Half-students-using-ChatGPT-cheat-rise-90.html

Fossen, F., & Sorgner, A. (2019). Mapping the Future of Occupations: Transformative and destructive effects of new digital technologies on jobs. Foresight and STI Governance, 13(2), 10–18. https://doi.org/10.17323/2500-2597.2019.2.10.18

Griggs, T., & Wakabayashi, D. (2018, March 20). How a self-driving uber killed a pedestrian in arizona. The New York Times. https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html

McDade, A. (2023, February 9). Studies show ChatGPT cheating is on the rise among students — young and old — as teachers remain divided on bans. Insider. https://www.businessinsider.com/teachers-caught-students-cheating-chatgpt-survey-shows-2023-2

Novet, J. (2023, January 18). Microsoft is laying off 10,000 employees. CNBC. https://www.cnbc.com/2023/01/18/microsoft-is-laying-off-10000-employees.html

Schank, R. C. (1987). What Is AI, Anyway?. AI Magazine, 8(4), 59. https://doi.org/10.1609/aimag.v8i4.623

Semenov, A. (2023, January 27). AI USE CASE: How a mobile game development studio saved $70K in expenses. Game World Observer. https://gameworldobserver.com/2023/01/27/ai-use-case-how-a-mobile-game-development-studio-saved-70k-in-expenses

Tsui, T. (2023, February 9). Coding with chatGPT – Tanya Tsui. Medium. https://medium.com/@tanyamarleytsui/coding-with-chatgpt-b50ab3fcb45f

Yadron, D., & Tynan, D. (2016, June 30). Tesla driver dies in first fatal crash while using autopilot mode. The Guardian. https://www.theguardian.com/technology/2016/jun/30/tesla-autopilot-death-self-driving-car-elon-musk