ARTIFICIAL INTELLIGENCE AND WORK

Autore: Dott. Enea Franza, Economist.

The application of artificial intelligence to the production sector[1] revives a long-discussed theme in human history, namely the conflict between work and technology. It should be noted, also to limit the field of analysis, that artificial intelligence (AI) is a part of computer science that aims to create systems that simulate or exhibit intelligent behavior.[2] AI algorithms use data and machine learning techniques to continuously improve their performance in specific tasks, such as image recognition, machine translation, or autonomous driving. Specialists distinguish between different types of AI, including weak AI (or basic AI) and strong AI (or general AI) where weak AI is designed to perform specific tasks, while strong AI is designed to exhibit human-like general intelligent behaviors.

In many industries, AI is already playing an important role in automating business processes, which is one of the most common applications in the industry. With the help of AI, companies automate production, logistics, and maintenance. For example, a manufacturing company can use an AI system to check the quality of products in real-time, thereby reducing human error and increasing efficiency. Additionally, AI can be used to optimize the operation of a manufacturing plant by automatically adapting processes based on external conditions in real-time. This can help reduce operational costs and improve resource utilization.

Beyond specific tasks that can be heavily automated, research conducted by McKinsey in the US estimates that the adoption of AI, alone, can bring greater growth to the entire US economy estimated at between 0.1 and 0.6% each year.

Artificial Intelligence, therefore, is already used today to automate tasks and solve complex problems, finding application in a wide variety of contexts: from scientific research to the stock market, from robotics to justice, passing through industry and self-driving cars[3].

But the fields of application are almost endless. In fact, artificial intelligence has long been adopted as a key technology alongside the IoT (Internet of Things[4]). A very common application is, for example, predictive maintenance. The analysis of the state of machinery, plants or equipment allows, using precise mathematical models, to prevent any failures, estimating the probability with which they could occur. Artificial intelligence is also particularly useful for personalizing products based on customer preferences. For example, a clothing company can use AI to analyze for each  target customer (youth, families, etc.) and create products tailored to meet their specific needs.

AI is also used to create demand forecasting models – for example, to identify purchasing patterns and market trends – to help companies plan production and avoid waste. AI can also be used for data analysis, helping businesses make more informed decisions. For example, a transportation company can use AI to analyze vehicle data and predict any mechanical problems before they occur. But there are so many fields in which it is used that it can be said without a shadow of contradiction that the impact far exceeds the economic field. And even the simple reconnaissance done allows us to understand how it lends itself to replacing human activity that had hitherto been employed to carry out tasks that, until recently, were carried out by man.

Justified fears, therefore, more than justified, also considering that there are many analyses that believe that no less than half of today's work activities could be automated by 2055. Therefore, it would seem that any type of work due to the pervasiveness of AI is abstractly subject to at least partial automation.

But the antagonism between capital and labor in the economic development of human civilization is by no means new. A first documented example can be found at the dawn of the first industrial revolution in England, where the antagonism between machines and work also resulted in a movement of struggle, known as Luddism, which carried out the intentional destruction of mechanical production facilities (machinery, factories) and which takes its name from Ned Ludd.  legendary leader of the protests that took place in England between 1811 and 1817. Luddism was a process in which the victims, who had unsuccessfully turned to authorities and manufacturers in order to prohibit the introduction of machines into industrial production, eventually organized themselves spontaneously and used force in order to oppose the presence of machines in factories.

But, let's take a step back and dwell on how equilibrium is achieved in the labor market and, to do so, let's first see how the amount of labor required by firms is determined and then analyze the supply. To answer the first seemingly simple question, we should make a number of considerations. Now, if it is assumed that the amount of labor that the firm will require will have to be such as to maximize profit, then the firm will use an additional worker if this will increase profits; In other words, the demand for labor of firms is a derivative demand, in the sense that firms determine the amount of desired product (which maximizes profits), and consequently determine the quantity of labor.

We know that the firm maximizes profits if the marginal cost equals the marginal revenue. An additional worker is an additional cost, but it also allows for higher production, and therefore additional revenue. Therefore, the amount of work will be such that a similar condition will be met. To determine the amount of labour required by a firm, we start from the amount of product that labour allows to generate. The marginal product of labour (in physical terms) is the increase in output that is achieved with an additional worker. The marginal product of labour in terms of value, on the other hand, is the increase in revenue due to the use of an additional worker.

Assuming a competitive market, a firm always sells its product at the same price, regardless of the quantity produced. The marginal product in terms of value indicates the additional revenue that the firm obtains by using an extra worker which, of course, for what has just been said will be given by the additional product of the worker multiplied by the selling price. On the other hand, the additional worker must be remunerated at the cost of an additional salary. Consequently, profits are greatest if the marginal product in terms of value is equal to wages.

We could then construct a curve that relates the money wage to the employed. The curve thus constructed can move in the Cartesian plane to the right or to the left, as other variables vary.

Now, with a simple example, we could show that an increase in the productivity of labour, i.e. the units produced for an additional unit of labour, and/or an increase in the selling price of the good produced, justifies an increase in the demand for labour by firms at the same wage per employee. On this point, we observe how an increase in productivity can be obtained using artificial intelligence, which at this point allows the same amount of work to obtain more product.

In the first table we have the starting situation, in the next one we verify that the introduction of an AI system the application determines an increase in the product.

TABLE 1

Now, if each worker's salary is 30,000 euros, the company will be able to hire two workers. As can be seen, in fact, the third worker has a cost higher than his marginal value and, therefore, it will not be profitable to hire an additional unit.

Let's go back to the situation where the unit price of the good sold is equal to 35 euros, and let's assume, instead, an increase in marginal productivity for each worker. We keep the cost of the salary constant, equal to 30,000 Euros per worker. Let the new situation be shown in the table below.

TABLE 2

In this case, it is profitable for the company to hire 4 units, as each worker makes more than it costs and, therefore, it follows that: "an increase in productivity will make it profitable to hire additional units of labour". 

Of course, to determine the actual amount of labour absorbed, the supply of labour must be considered. It is certainly decided by wages, in the sense that an increase in wages may correspond to an increase in the supply of people willing to work, but other variables certainly condition supply. However, the above example may be useful for us to determine in the abstract the effect of an increase in productivity.

Coming back to us, the ongoing debate on the topic of AI branches off into two branches: the one concerning the relationship between technology and jobs, which focuses on the impact of AI on wages, labour demand and the creation of new jobs, and which sees a school of pessimists confront a large group of scholars convinced of the benefits in terms of productivity of AI,  and the one that focuses on the technology-work aspect, that is, that analyses the implications of AI for autonomy and quality of work. Well, the effects of AI on the compression of demand and pay for labour exceed those posted. As an automation technology, in fact, AI would be intrinsically capable of expanding the set of functions within the production process performed by capital rather than by labour and in this way would implement a progressive but constant substitution effect no longer on manual work but also on important shares of cognitive work related to repetitive tasks.

Now the aforementioned substitution effect would trigger a spiral of reduction in the supply of labour and lower wages, which would also be followed by a crisis in access to consumption. The increase in productivity would therefore coincide with a decoupling between the accumulation of capital and the increase in the profitability of labour[5]. On closer inspection, the transition to a capital-intensive economy to the detriment of one with high labour absorption would seem to be confirmed by isolating the experience of the manufacturing sector in recent years. In fact, about 50 years ago, the most important manufacturing company produced cars and was General Motors; it generated, before the crash of the 1980s, about $50 billion in profits each year and employed over 800,000 people. Today, the most important manufacturing company is Apple and produces operating systems, smartphones, computers and multimedia devices, generating about 100  billion in profit by employing about 137,000 full-time employees in 2020.

Now, the World Economic Forum[6] itself calculates that, by 2025, about 85 million jobs will be replaced by artificial intelligence.

However, in the same report, it is stated that EUR 97 million will be created and that, therefore, the balance sheet will ultimately be extremely positive in terms of employment. In addition, studies find that 5.5% of total employment in high-income countries is potentially exposed to the automation effects of technology, while in low-income countries the risk of automation affects only about 0.4% of employment. McKinsey in 2018 estimated that AI will be responsible for an increase in world GDP of 1.2% per year, three times what robotization produced during the 90s and twice as much as it was achieved thanks to the spread of information technology in the 2000s.[7]

Today, Artificial Intelligence is used to automate tasks and solve complex problems, finding application in a wide variety of contexts: from scientific research to the stock market, from robotics to justice, passing through industry and self-driving cars. In Italy, a recent study conducted by EY, Manpower Group and Sanoma Italia[8], estimates that the demand for labour will increase in the next 10 years due to AI, in 9 out of 23 sectors of activity and, among these, also in some technologically mature sectors (telecommunications, public utilities, chemicals), and in sectors related to the transformation of services and skills (care services,  education, training and employment services). Among the sectors in which aggregate labour demand is expected to decline are banking and insurance, which have long embarked on a restructuring path linked to the use of data technologies. According to the survey, between now and 2030, the demand for technical and high-skilled professions will increase more and more, not only related to IT and technology, but also to care and services related to people, including guidance, training and socio-work integration. On the other hand, demand will fall for lower-skilled occupational groups, as well as for skilled and entrepreneurial professions linked to low-growth sectors (e.g. primary sector, traditional industries).[9] Here, too, the forecasts reported in the aforementioned report believe that, overall, the demand for labour in Italy will remain growing for the rest of the decade.

That said, ultimately, therefore, AI should work by automating some tasks rather than completely replacing them by making some activities carried out by humans anachronistic, especially those that are more dangerous, risky or repetitive. This, on closer inspection, would allow you to focus on professional, recreational, cultural and socially useful areas, experimenting with unexplored novelties. From this point of view, the jobs that would be created would be even more numerous than those lost.  

In other words, likely, Artificial Intelligence will take care of tasks that require tasks such as copying, pasting, transcribing, and typing and, in any case, will be used in production to make processes more efficient.

If what has been analysed and suggested is true, the question becomes, not so much about whether AI will take away work, but how it will change the world of work itself,[10] which it would seem to understand would find ample lymph in the most creative jobs and require, of course, new skills. Soon, for example, AI-assisted health technician jobs will see an upward surge. Artificial Intelligence, moreover, is already playing an important role in the automated transport sector. For some time now, companies such as Uber and Google have invested (with very questionable results) significant amounts in self-driving cars and trucks and, as this mode of transport resumes, most likely soon, it will create many jobs for specialized engineers[11]. This analysis, of course, applies to all sectors, where the demand for personnel responsible for updating, managing, and maintaining AI tends to increase steadily. Companies would need large amounts of developers, for example, to manage their systems.

If this interpretation is the one that most closely captures the essence of the phenomenon, the "politics of knowledge" appears to be the key that would make it possible to govern the negative structural effect, as the workforce would be assigned the role of directing the technological process. In fact, it is only by intervening in this context that it is possible to counteract the phenomenon of "mismatch", an event that is revealed, precisely when, although there is a strong demand for professionals, the labour market is nevertheless unable to provide enough qualified candidates for the specific job positions. On this point, recent data relating to our country[12] see the share of hires that Italian companies consider difficult to achieve exceed 48% in September 2023 (a figure that has been growing continuously since at least 2019), while the percentage of jobs available, but not occupied is around 2%, with estimated losses of 3% of the annual added value of industry and services.

But already now, those who possess specific skills in terms of artificial intelligence, even limited to the ability to use the various tools on the market, seem to have more opportunities to be hired, as confirmed by the Linkedin report "Future of world Report". AI at work" of 2023, which notes that between November 2022 and June 2023, job offers on the portal requiring knowledge of GPT or Chat GPT have significantly increased, although without being accompanied by the possession of a specialist degree in the field, and thus proving, in this respect, a discrepancy between digital skills possessed and degree qualifications.

The remedy to the talent shortage and mismatch is given, then, by training.

It would seem to be quite evident that technological development will require a greater need for AI-related skills in the near future, to the point that even simply by integrating AI into learning processes it may be easier and faster to align the offers of education systems with the constant transformations of the labour market. A fundamental role will have to be played by guidance already in secondary schools, which to achieve the result, will have to be reset in order to allow students and families to focus on the acquisition of skills and to recognize which training paths and which professional choices offer greater opportunities for success.

On the other hand, according to various analysts, artificial intelligence[13] will be an increasingly valuable and effective resource in the training offer in the future, making courses and training programs more accessible for workers and companies, as well as allowing an increase in the effectiveness of higher education by enhancing traditional training solutions.

[1] According to the traditional classification, the economy consists of three sectors. The primary sector corresponds to basic production: agriculture and livestock; forestry (forestry), fisheries and extractive industries; The secondary sector concerns the production of goods: energy, industry, construction, crafts and the tertiary sector which includes everything related to services, i.e., trade, tourism, sport, entertainment, social services (education, health, justice, public administration), economic and financial activities (banks, stock exchanges, insurance), transport, scientific research. However, the development of technology has made it necessary to define an additional sector that deals with particular services, the so-called advanced tertiary or quaternary sector, which is linked to information and telematic services.

[2] According to  the definition given by the European Union, artificial intelligence (AI) is "the ability of a machine to display human abilities such as reasoning, learning, planning, and creativity." In other words, a set of technologies that allows systems to "read" their environment through data, relate to what they perceive, solve problems and act towards a specific goal based on specific algorithms.

[3] What emerges from these first studies is that the use of AI certainly brings a significant increase in the productivity of the company, but the effect unfolds very differently depending on the reference sector, with an increase in the effects on productivity in highly repetitive office tasks. For example, according to a study conducted by the Nelsen Norman Group, a company specializing in the interfaces of technological products, the massive use of AI in customer service activities has led to an increase in the speed of responding to consumers estimated in the order of more than 10%. On the other hand, when it comes to the drafting of routine business documents, office productivity has increased by almost 60%. Furthermore, the capacity of IT companies that develop software with little innovative content has increased to levels of 120%.

[4] The Internet of Things refers to the  technological development according to which, through the Internet, every object acquires its own identity in the digital world. As mentioned, therefore, IoT is based on the idea of "smart" objects interconnected with each other in order to exchange the information owned, collected and/or processed.

[5] In this regard, Gordon's (2018) studies are fundamental, comparing data on productivity in the manufacturing sector among OECD countries, arguing that AI fully confirms the paradox of productivity: despite expectations and what has happened in the past, AI would not have the ability to significantly affect productivity. This is demonstrated by the fact that in the last decade considered in the study (2006-2016) productivity, i.e. the value added produced per worker, would have been lower overall in the manufacturing sector than in the previous one.

[6] "The Future of Jobs Report 2020. October 2020". The Report identifies a number of jobs in rapid employment growth (Data Analysts and Scientists 2 AI and Machine Learning Specialists 3 Big Data Specialists 4 Digital Marketing and Strategy Specialists 5 Process Automation Specialists 6 Business Development Professionals 7 Digital Transformation Specialists 8 Information Security Analysts 9 Software and Applications Developers 10 Internet of Things Specialists 11 Project Managers 12 Business Services and Administration Managers 13 Database and Network Professionals 14 Robotics Engineers 15 Strategic Advisors 16 Management and Organization Analysts 17 FinTech Engineers 18 Mechanics and Machinery Repairers 19 Organizational Development Specialists 20 Risk Management Specialists) and compares them with those in equally rapid decline (Data Entry Clerks 2 Administrative and Executive Secretaries 3 Accounting, Bookkeeping and Payroll Clerks 4 Accountants and Auditors 5 Assembly and Factory Workers 6 Business Services and Administration Managers 7 Client Information and Customer Service Workers 8 General and Operations Managers 9 Mechanics and Machinery Repairers 10 Material-Recording and Stock-Keeping Clerks 11 Financial Analysts 12 Postal Service Clerks 13 Sales Rep., Wholesale and Manuf., Tech. and Sci. Products 14 Relationship Managers 15 Bank Tellers and Related Clerks 16 Door-To-Door Sales,  News and Street Vendors 17 Electronics and Telecoms Installers and Repairers 18 Human Resources Specialists 19 Training and Development Specialists 20 Construction Laborers).

[7] McKinsey (2018), Accenture (2017) and PWC (2018). Accenture in its "Reworking the revolution: are you ready to compete as intelligent technology meets human ingenuity to create the future workforce" released on the occasion of the Davos Economic Forum estimated that companies' revenues could grow by 38% by 2020, provided they invested in Artificial Intelligence and effective human-machine cooperation.

[8] "The Future of Skills in the Age of Artificial Intelligence," Predictive Study 2023, https://competenze2030.it/. The study, in particular, notes that the growth in AI-related demand will concern very heterogeneous profiles: engineers and physicists (+7%), but also market analysts and occupational and training psychologists (+3%). There will be an increase in demand for highly creative profiles (architects, designers, planners), but also for professions related to marketing and sales (+5%). The impact of AI on the reorganisation of work processes and models will be evident in the growth in demand for managerial professions, such as directors of administration and finance and organisation specialists (+3%). 

[9] Based on the survey on companies and work, conducted by INAPP in 2022, it emerges that 1.2% of the total number of Italian companies have invested in AI, estimating that these investments are associated with a significant increase in the propensity to search for highly qualified professional profiles (+2%), while exerting no effect on the demand for medium-skilled or unskilled professions.

[10] A new study by the International Labour Organization (ILO) also supports the optimists: "Generative AI and Jobs: A global analysis of potential effects on job quantity and quality", August 2023.

[11] Incidentally, it should be noted that 2022 was the year that saw the closure of the most companies and start-ups dedicated to autonomous driving, while 2023 was characterized by the stop of Cruise's activities, with its robo-taxis abandoning American roads. Even less advanced systems, such as Tesla's Autopilot driver assistance system, lend themselves to numerous criticisms.

[12] INDIRE, 2022 national monitoring data, presented on 14 June at the Ministry of Education, at the "ITS DAY", an event dedicated to professional tertiary training.

[13] See "The Future of Skills in the Age of Artificial Intelligence," 2023 Predictive Study, https://competenze2030.it/.