Abstract
This publication represents the third of a cycle of analyses exploring the geopolitical consequences of Artificial Intelligence (AI). The global integration of Artificial Intelligence (AI) represents a dynamic shift across the economic and financial domains, posing new challenges to regulators. In the economy, AI fosters productivity and decision-making enhancements, leading to substantial growth projections. In finance, the newest technologies have revolutionized credit risk modeling, fraud detection, and portfolio optimization by enhancing the professionals’ capabilities of acting on and around the markets. Nonetheless, rising concerns persist about job displacement, economic disparities, and ethical considerations. The pressing need for international regulations emerges to navigate AI's multifaceted impacts, ensuring a balanced, secure, and ethically-driven global economy.
Author
Michele Gioculano - Head Researcher, Mondo Internazionale G.E.O. - Politics
Lorenzo Molina - Junior Researcher, Mondo Internazionale G.E.O. - Politics
Simone Mezzabotta - Junior Researcher, Mondo Internazionale G.E.O. - Politics
Introduction
Artificial Intelligence (AI) has become an integral force shaping the economic and financial landscape, ushering in transformative changes and presenting novel challenges for regulators. As AI's influence extends across sectors, it has the potential to drive economic growth and revolutionize decision-making processes. However, the global adoption of AI raises pertinent questions about the need for international regulations to navigate its multifaceted impacts. This is crucial for ensuring a balanced, secure, and ethically-driven global economy.
In the following sections, we explore the economic implications of AI, focusing on its potential to drive productivity gains, create new markets, and reshape labor dynamics. Subsequently, the paper delves into the transformative role of AI in finance, elucidating how Machine Learning (ML), Deep Learning (DL), and Natural Language Processing (NPL) have become indispensable tools in risk assessment, fraud detection, and investment optimization. The final section underscores the necessity of international regulations, examining how legislative frameworks can address concerns related to workforce displacement, economic inequality, and the ethical use of AI.
As the international community grapples with the regulation of AI, recent initiatives, such as those undertaken by the European Parliament, highlight the urgency of comprehensive and forward-thinking regulatory frameworks. The dynamic nature of AI demands proactive measures to ensure that its potential benefits are harnessed responsibly, mitigating risks and fostering a harmonious integration of this technology into the global economic and financial systems.
Economy and AI
Artificial intelligence is assuming an increasingly significant role in both our daily lives and the economy, already exerting a substantial influence on various aspects of our world. The global competition to harness its advantages has become intense, with prominent players like the US and China taking the lead. Regarded by many as a catalyst for productivity and economic expansion, AI has the potential to enhance operational efficiency and revolutionize decision-making processes through the analysis of extensive datasets. Furthermore, it holds the capacity to stimulate the creation of novel products, services, markets, and industries, thereby fostering consumer demand and generating fresh revenue streams (Szczepański, 2019).
A study conducted by PricewaterhouseCoopers (PwC) in 2018 suggested that the global Gross Domestic Product (GDP) could experience a substantial increase of up to 14% by 2030, equivalent to US$15.7 trillion, owing to the rapid development and widespread adoption of Artificial Intelligence. The report envisioned a forthcoming wave of the digital revolution, propelled by the data generated through the Internet of Things (IoT), surpassing the magnitude of data generated by the current 'Internet of People.' This surge is expected to drive standardization, foster automation, and concurrently, enhance the customization of products and services (PwC, 2018).
PwC identified two primary channels through which AI is poised to impact the global economy. Firstly, AI is anticipated to deliver productivity gains in the short term by automating routine tasks, particularly in capital-intensive sectors such as manufacturing and transportation. This automation may encompass the extended use of technologies like robots and autonomous vehicles. Additionally, businesses are expected to enhance productivity by integrating AI technologies to complement and assist their existing workforce. This involves investments in software, systems, and machines that leverage assisted, autonomous, and augmented intelligence. Such investments would not only empower the workforce to execute tasks more effectively but also liberate time for employees to concentrate on more stimulating and higher value-added activities. Consequently, automation is projected to reduce the reliance on labor input, resulting in overall productivity gains.
Over time, the second channel — the availability of personalized and higher-quality AI-enhanced products and services — will assume even greater significance. This accessibility is poised to stimulate consumer demand, creating a positive feedback loop wherein increased consumption generates more data. PwC articulated this process by stating that "increased consumption creates a virtuous cycle of more data touchpoints and hence more data, better insights, better products and hence more consumption." While the benefits will extend globally, the US and China are anticipated to reap the most substantial rewards from AI technology. In particular, the US is poised to swiftly introduce numerous productive technologies, with gains accelerated by advanced AI readiness among businesses and consumers, a rapid accumulation of data, and heightened customer insight (PwC, 2018).
Moreover, according to Goldman Sachs Research, breakthroughs in generative artificial intelligence have the potential to bring about sweeping changes to the global economy. As evidenced by Goldman Sachs economists Joseph Briggs and Devesh Kodnan, in the face of considerable uncertainty regarding the capabilities of generative AI, its ability to generate content that is indistinguishable from human-created output and to break down communication barriers between humans and machines signifies a significant breakthrough, carrying the potential for substantial macroeconomic impacts. According to them, the advent of new systems is poised to exert a considerable influence on global employment markets. The transformative changes in work processes brought about by these advancements may render around 300 million full-time jobs susceptible to automation. Taking the example of the USA, upon scrutinizing databases containing task specification for over 900 occupations, they projected that approximately two-thirds of U.S. occupations face some level of exposure to AI-driven automation. Their estimates suggest that, within these exposed occupations, between a quarter and half of their tasks could be replaced by automation. However, they emphasized that not all tasks automated by AI will result in job layoffs. While the impact on the labor market is anticipated to be substantial, most jobs and industries are only partially exposed to automation, making them more likely to be complemented rather than entirely substituted by AI.
However, the widespread adoption of AI may also introduce profound disruptions to the economy and society. Some express concerns about the emergence of super firms, concentrated hubs of wealth and knowledge, that could adversely impact the broader economy. There is also apprehension regarding the potential exacerbation of the developmental gap between affluent and developing nations, along with a shift in demand for specific skill sets that may render certain jobs obsolete, significantly impacting the labor market. Experts additionally caution against the possibility of heightened inequality, reduced wages, and a contraction of the tax base.
The transformative impact of AI is poised to extend beyond technological realms, significantly influencing wage structures, income distribution, and economic equality. The burgeoning demand for highly skilled workers adept at leveraging AI capabilities may propel an upward trajectory in their wages. In contrast, a considerable portion of the workforce may encounter a wage conundrum or potential unemployment.
This divergence in wages could have ripple effects, affecting even mid-skilled workers whose earning potential might face downward pressure. The utilization of AI by high-skilled workers not only enhances their productivity but also broadens the spectrum of tasks they can efficiently undertake. Consequently, the evolving labor dynamics may exacerbate overall income inequality by influencing wage patterns across the board (Szczepański, 2019).
The extent of these repercussions will hinge on the pace of AI adoption, with rapid changes likely to yield more adverse effects, given existing market imperfections. Theoretically, as AI progressively replaces routine labor, overall productivity and income growth may surge, simultaneously intensifying economic inequality. This scenario may give rise to a 'paradox of plenty,' wherein society experiences heightened overall prosperity. However, for many individuals, communities, and regions, technological advancements could inadvertently reinforce existing inequalities (ibidem).
The Transformative Role of AI in Finance
Among the sectors in which AI has emerged as a pivotal force, finance is arguably one of the most concerned. Its integration into financial systems has unlocked unprecedented opportunities, revolutionizing traditional practices and optimizing decision-making processes. At the same time, future outlooks seem inevitably intertwined with challenges, ethical considerations, and regulation developments. Broadly speaking, AI in finance encompasses a spectrum of technologies that are generally used for different purposes, including Machine Learning, Deep Learning, and Natural Language Processing. Although the terms AI and machine learning are frequently used interchangeably, their scope differs: while all types of machine learning fall under the umbrella of AI, not all AI systems rely on machine learning. AI encompasses a broader concept aiming to develop machines that emulate human cognition. In contrast, machine learning represents a subset of AI enabling machines to learn from data without requiring explicit reprogramming. Additionally, within machine learning exists a specialized subset known as deep learning, which employs algorithms and architectures inspired by the human brain.
Machine Learning (ML), a subset of AI, equips financial institutions with the ability to learn from data patterns and make predictions or decisions without explicit programming. More specifically, ML algorithms are extensively used for risk assessment, which is based on historical data and enables more accurate predictions. Since the early 2000s, there has been extensive academic exploration regarding the utilization of machine learning methodologies for credit risk modeling. For instance, Angelini et al. (2007) employed a neural network approach to model credit risk among Italian SMEs, albeit on a limited dataset. Similarly, Auria and Moro (2008) investigated company solvency utilizing support vector machines, noting superior out-of-sample predictive accuracy compared to established methods. Furthermore, Khandani et al. (2010) utilized generalized classification and regression trees (CART) on a substantial dataset from a commercial bank. Their model integrated conventional credit metrics like debt-to-income ratios with consumer banking transactions, significantly augmenting predictive capabilities.
ML algorithms have also proven valuable in fraud detection. Recently, specialists from Allianz UK have collaborated on a novel machine-learning tool aimed at tackling the rise in fraudulent claims: ‘Incognito’. Incognito is reportedly able to identify suspicious claims and refers them to fraud specialists for thorough review and investigation. Allianz disclosed that the utilization of Incognito in identifying fraudulent claims has resulted in savings of £1.7 million thus far (Insurance Business 2023). Additionally, an extra £3.4 million remains held in claim reserves pending the completion of ongoing investigations (Insurance Business 2023). Furthermore, MasterCard has leveraged AI and ML solutions provided by AWS to enhance its worldwide fraud detection capabilities for various years. Currently, the credit card behemoth is also collaborating with Persistent to explore methods that accelerate the deployment of cloud-based software, aiming to optimize operational efficiency and agility (SiliconAngle 2023).
Finally, machine learning techniques are primarily responsible for popularizing algorithmic trading, which requires a real-time analysis of market data to execute trades based on predefined parameters. Recently, algorithmic trading has become a prevalent practice among traders and investors in India, enabling swifter and more efficient deal executions (Financial Express 2023). This method allows them to seize market opportunities that might otherwise be arduous to uncover through manual processes. The Indian stock market is, in fact, recognized for its volatility, dynamism, and nonlinear nature, influenced by diverse factors like geopolitical shifts, unforeseen events, and economic conditions. These multifaceted dynamics make it a challenging environment to navigate and maintain stability. Nonetheless, the rise of algo trading has transformed the strategies employed by traders and investors in approaching this intricate market landscape and de facto revolutionized the market landscape, offering the potential to forecast stock trends and behaviors more accurately (Financial Express 2023).
Deep Learning (DL) focuses on neural networks capable of learning from vast amounts of unstructured data. Despite being a subset of ML, its specific use cases in finance suggest a separation of the two for a clearer understanding of the impact of AI in the sector. Firstly, DL techniques are instrumental in predictive analytics, which requires processing complex financial data - e.g., market trends, sentiment analysis, and economic indicators - to predict future market behavior more accurately. Secondly, neural networks may be used to optimize investment portfolios by analyzing diverse asset classes, adjusting risk levels, and maximizing returns based on dynamic market conditions. Although the Markowitz model has laid the groundwork for contemporary investment strategies in finance, the challenges persist in seeking time-efficient portfolio analysis, particularly in the realm of high-frequency trading (HFT), where dynamic stock price fluctuations complicate risk-return objectives. Hence, a recent study introduced a novel solution utilizing a recurrent neural network (RNN) model to address real-time portfolio optimization challenges. Employing empirical data from Dow Jones Industrial Average (DJIA) components, numerical experiments showcased the superiority of the proposed RNN model over the DJIA index, yielding higher investment returns and mitigated risks (Cao et al. 2023) Finally, DL models can enhance credit scoring systems by evaluating multiple data points, leading to more comprehensive and accurate assessments of creditworthiness. For instance, Wang and Xiao (2022) pioneered the integration of a transformer architecture into the domain of credit scoring by leveraging user online behavioral data. Empirical evaluations demonstrate the superiority of the proposed Feature Embedded Transformer (FE-Transformer) deep learning model over other comparative methods, underscoring the robust predictive capability of the FE-Transformer deep learning model in accurately assessing user default risk within credit scoring systems in finance (Wang & Ciao 2022).
Finally, Natural Language Processing (NLP) tries to enable computers to understand, interpret, and generate human language. More specifically, NLP unites computational linguistics, which involves rule-based representation of human language, with statistical, machine learning, and deep learning models. These amalgamated technologies empower computers to analyze and interpret human language, encompassing both textual and vocal data, comprehending its nuanced meaning, and discerning the intentions and sentiments of the speaker or writer (IBM 2023). In finance, NLP facilitates sentiment analysis by analyzing textual data to gauge market sentiment and provide insights into investor behavior and market trends. For instance, Yadav et al. (2019) tried to propose a ‘semantic orientation based unsupervised approach for finding sentiments strength of financial text’ (Yadav et al. 2019, p. 1). Their findings suggested that the so-called ‘noun-verb approach’ emerged as the most effective tool, but, in any case, sentiment analysis seemed to accurately gauge investors’ attitudes and predict stock fluctuations (Yadav et al. 2019). Furthermore, NLP models usually power chatbots and virtual assistants that can interact with customers, handle inquiries, provide financial advice, and assist in various banking operations. Finally, NLP tools assist in parsing and understanding regulatory documents and updates, ensuring financial institutions remain compliant with evolving regulations.
In conclusion, finance stands as a sector significantly impacted by the rise of AI, offering unparalleled advancements, reshaping conventional methods, and enhancing decision-making processes. As AI penetrates financial systems, it introduces transformative opportunities while also raising concerns about ethics, regulatory challenges, and future developments.
International regulations
As has been widely described previously, artificial intelligence plays and, most likely, will play a very important role in the economic-financial field. Sectors that are particularly vital for a country and closely connected not only to its stability and international strength but, more simply, to the daily life and needs of its citizens. Therefore, although artificial intelligence is currently little more than a newborn, the need to develop and impose regulations on this new form of technology was immediately placed at the center of the debate on the topic. In addition to being a prerogative of modern states and complex societies, the need to legally frame such an area derives from the evident rapidity of its expansion and the unknown goals, positive or otherwise, to which it could lead in the short term. Consequently, since legislative times are much longer than those demonstrated by technological progress and artificial intelligence is a particularly vast field, many institutions, both public and private, are already working to provide adequate regulation.
Making specific reference to the economy, artificial intelligence could give rise to a new digital revolution, leading to a progressive marginalization of human contribution and, ultimately, to a drastic drop in the demand for workforce with very significant effects on entire communities. This effect would risk being particularly disruptive, especially in the West, where the leading economic sector is no longer represented by agriculture or industry but by services, where artificial intelligence appears to be more successful and find better application. On the contrary, emerging or developing economies, which do not boast very evolved and technologically advanced economic systems, would suffer less from the marginalization of the workforce, requiring more time to progress and integrate such advanced technologies into their processes. Consequently, one of the necessary legislative interventions to be considered and implemented is, without a doubt, represented by a system of protection of human capital and the contribution of the worker, adequately balancing the inevitable technological evolution and the useful contribution provided by artificial intelligence with the maintenance of employment levels appropriate to the needs of a community.
Another central aspect of the issue is represented by security. In fact, artificial intelligence, if used for illicit purposes, could become a formidable tool in the hands of criminals. Both small-time criminals and large unscrupulous fixers and speculators, using the means that these advanced technologies make available, could deceive customers and engage in unfair competition, to the point of manipulating the market for their own gain. The aforementioned cases would risk undermining an economic-financial system from within or ensuring, for individuals or companies, a preponderant position within the market and an excessively large and conditioning power over entire communities. Therefore, it is extremely essential to provide very strict control measures on the use of artificial intelligence and its application in certain contexts and for certain purposes. It is also important to place effective limits on the possibility of concentration of economic and technological capabilities, in the field of artificial intelligence, in the hands of a few people or companies which would risk giving rise to dominant and difficult-to-control monopolies or oligopolies.
As is known, the European Parliament is preparing to approve a framework agreement that regulates the vast and complex subject of artificial intelligence. The provisional agreement, reached on December 8, after fifteen hours of negotiation, is aimed at regulating the use of artificial intelligence by governments for biometric surveillance and the use of ChatGPT. Numerous technical support documents that are in favor to imposing regulations for such advanced technology will now need to be agreed upon and drafted. Equal importance will be given to cyber security and the environmental impact of artificial intelligence, a source of significant energy consumption. Attention to privacy and security constitutes, in any case, the most important aspect, a potential starting point for future more specific regulations focused on the various application sectors. Although Europe is not always brilliant in terms of speed of action, it is possible to argue that, in this case, Brussels is a true pioneer on a global level. Since, by nature, the law follows the need, only time and the large-scale use of all the potential of artificial intelligence will make it possible to analyze and regulate its various facets for the benefit and safety of all.
Conclusions
This paper analyzed the multiple applications of artificial intelligence and its impacts, both on the economic and financial fronts. Since this is a technology in continuous and constant evolution, many possibilities still have to be applied or verified but there is no doubt that its potential is already particularly significant today. Like these, any risks and negative implications could also be very important. For this reason, the need has been highlighted to take action, starting now, given the long delays of the legislative apparatus, to build effective regulation and control systems aimed at supervising sectors and methods of application of artificial intelligence as well as framing, from a legal point of view, the entire matter. It was then observed how even more important and effective it would be to reach a broad international agreement functional to the drafting of common legislation that can, with its broad scope, preserve high safety standards at a global level.
Furthermore, it is important to remember that, with respect to the application of artificial intelligence to the economic-financial sectors, an ethical debate also occurs. The adoption of highly advanced technologies, aimed at maximizing profits, which do not take into account the needs of the human being and which, on the contrary, risk marginalizing his role compared to the machine certainly requires profound reflection. Its exclusion from the production process, even in those creative and intellectual sectors in which, even after the industrial and information revolution, it seemed destined to maintain an undisputed predominance opens up unexplored and, perhaps, risky scenarios, in an era of transition like this. Therefore, international regulators will be forced to deal with this issue too.
However, in a period of crisis of multilateralism and global cooperation, the most recent signals, such as the regulatory initiatives of the European Parliament and the London Summit, seem to suggest that the major Powers have understood both the potential and the risks of artificial intelligence and that they are adequately taking action not only to benefit from the economic and financial benefits of this new technology but also to guarantee its regular and virtuous development.
In the next publication of this cycle, will be analyzed the geopolitical implications of AI on International Relations.
Classification of sources and information:
1 | Confirmed | Confirmed by other independent sources; logical in itself; coherent with other information on the topic |
2 | Presumably true | Not confirmed; logical in itself; coherent with other information on the topic |
3 | Maybe true | Not confirmed; reasonably logical in itself; coherent with some other information on the topic |
4 | Uncertain | Not confirmed; possible but not logical in itself; no other information on the topic |
5 | Improbable | Not confirmed; not logical in itself; contradicts with other information on the topic |
6 | Not able to be evaluated | No basis to evaluate the validity of the information |
Trustworthiness of the source | ||
A | Trustworthy | No doubt about authenticity, reliability or competence; has a history of total trustworthiness |
B | Normally trustworthy | Small doubts about authenticity, reliability or competence, nevertheless has a history of valid information in a majority of cases |
C | Sufficiently trustworthy | Doubts about authenticity, reliability or competence; however, has supplied valid information in the past |
D | Normally not trustworthy | Significant doubt about authenticity, reliability or competence, however has supplied valid information in the past |
E | Not trustworthy | Lack of authenticity, reliability or competence; history of invalid information |
F | Not able to be evaluated | No basis to evaluate the validity of the information |
Bibliography:
- Anita Yadav et al. (2019). International Conference on Computational Intelligence and Data Science (ICCIDS 2019). Procedia Computer Science Volume 167, 2020, Pages 589-598.https://www.sciencedirect.com/science/article/pii/S1877050920307912. 1-A
- Bart van Liebergen (2023). Machine Learning: A Revolution in Risk Management and Compliance?. THE CAPCO INSTITUTE JOURNAL OF FINANCIAL TRANSFORMATION. https://www.iif.com/portals/0/Files/private/32370132_van_liebergen_-_machine_learning_in_compliance_risk_management.pdf 1-A
- Chongren Wang & Zhuoyi Xiao. (2022). ‘A Deep Learning Approach for Credit Scoring Using Feature Embedded Transformer’. MDPI. Appl. Sci. 2022, 12(21), 10995. https://www.mdpi.com/2076-3417/12/21/10995. 1-A
- FE TransformX (2023). ‘Machine Learning and AI in Algo trading: Predictive analytics for stocks.‘ Available at https://www.financialexpress.com/business/digital-transformation-machine-learning-and-ai-in-algo-trading-predictive-analytics-for-stocks-3244943/. Accessed on 13/12/2023. 1-B
- Foo Yun Chee, Martin Coulter & Supantha Mukherjee, ‘Europe agrees landmark AI regulation deal’, Reuters, 11 December 2023, https://www.reuters.com/technology/stalled-eu-ai-act-talks-set-resume-2023-12-08/ 1-B
- Goldman Sachs (2023). ‘Generative AI could raise global GDP by 7%’. Available at https://www.goldmansachs.com/intelligence/pages/generative-ai-could-raise-global-gdp-by-7-percent.html 1-B
- IBM. Natural language processing. IBM. Retrieved from https://www.ibm.com/topics/natural-language-processing#:~:text=Natural%20language%20processing%20(NLP)%20refers,same%20way%20human%20beings%20can (Accessed on 10/12/2023) 2-B
- Marcin Szczepański (2019). ‘Economic impacts of artificial intelligence (AI)’. European Parliamentary Research Service. 1-A
- Mariano-Florentino Cuéllar, ‘The UK AI Safety Summit Opened a New Chapter in AI Diplomacy’, Carnegie Endowment for International Peace, 9 Novembre 2023, https://carnegieendowment.org/2023/11/09/uk-ai-safety-summit-opened-new-chapter-in-ai-diplomacy-pub-90968 1-B
- PwC (2018). ‘The macroeconomic impact of artificial intelligence’. 1-B
- Silicon Angle (2023). ‘MasterCard extends its fraud detection capabilities through partnership with Persistent Systems and AWS‘. Available at https://siliconangle.com/2023/12/08/mastercard-extends-its-fraud-detection-capabilities-through-partnership-with-persistent-systems-and-aws-supercloud5/. Accessed on 11/12/2023 1-A
- Xinwei Cao et al. (2023) Expert Systems with Applications. Volume 233, 15 December 2023, 210934. Available at https://www.sciencedirect.com/science/article/pii/S0957417423014367. 1-A