The risks and challenges of using AI in the professional context

It is essential to highlight the risks associated with the introduction of AI into the working environment, namely :

1. misinformation, which provides misleading information to the various players and distorts results, leading to a loss of connection with the reality of the work itself ;
2. AI discrimination and toxicity, where the creation and use of discriminatory and toxic information can have legal consequences, given its link to ethical issues;
3. malicious actors and the abuse of AI, given that there are yet no adequate regulations to control the use of AI within companies;
4. new rules and restrictions arising from “man-machine” interactions resulting from the use of AI;
5. the additional threats to privacy and the security of information systems that AI brings ;
6. the new socio-economic and environmental damage caused by the adoption of AI; and finally,
7. the safety, bugs; and limitations of AI systems (AI Risk Repository, 2024).

One of the specific risks associated with the introduction of AI is the impact of its interface on user interaction. If there is a risk related to the psychological reactions of users when interacting with AI systems, companies and their employees may become overly dependent on this technology, which could lead to a loss of autonomy and an overreliance on AI, potentially compromising human intervention and judgment (AI Risk Repository, 2024).

In addition to the risks associated with misuse and malicious actors, those related to a lack of privacy and information systems security are the responsibility of companies.

To meet these challenges, regulations and standards such as the GDPR (General Data Protection Regulation), NIS 2 (Network and Information Security 2), and ISO 27001 (Information Security Management System) have been developed to help organizations strengthen information protection.

However, as mentioned above, AI can also cause socio-economic damage, either indirectly, by amplifying other risks, or directly, when it is used to centralize corporate decision-making power and profits. Furthermore, this risk materializes when AI contributes to dehumanizing work, reduces job quality, eliminates jobs, or fosters new forms of discrimination between users and designers. Finally, while data centres and AI offer simplicity and efficiency in their use, they represent a significant environmental risk due to their high energy consumption (AI Risk Repository, 2024).

As mentioned above, data biases represent one of the most important ethical challenges facing AI. Algorithms are often influenced by biases inherent in the data used, and the primary challenge is to ensure that these biases are addressed neutrally and fairly, thereby avoiding unwarranted negative or positive consequences (Baeza, 2021, cited in González, 2024).

Massive data, commonly referred to as Big Data, has generated significant disruptions in terms of privacy and security during the processing of personal data. It is therefore proposed that governments establish ethical and legal frameworks that ensure both the well-being of the population and the protection of their personal information (Arrigain, 2020).

In this context, it’s not just people who are aware of the use of AI technologies who are concerned, but also vulnerable and easily manipulated groups, such as children and older people. In addition, large companies profit and enrich themselves by collecting data generated at no cost to users (Coeckelbergh, 2020).

AI risks in the work context: Automation is one of the most significant risks associated with using AI in work contexts, as it can lead to structural unemployment, particularly in routine and repetitive tasks. For example, Foxconn cut 60,000 jobs after introducing robots into its processes (Corvalán, 2019, p. 38). In the same vein, Goñi (2019) notes that mass automation could reduce human interaction in the workplace and impact its quality, potentially leading to the dehumanization of employment.

Another risk associated with Artificial Intelligence in the workplace is the inequality and polarization of the workforce, which affects the demand for medium-skilled workers. As automation progresses, there is an increase in the hiring of high-skilled employees for specialized tasks and low-skilled employees for less complex jobs, widening the wage gap and generating an unequal distribution of opportunities in the labor market (Bravo et al., 2018, pp. 13-14).

In the same sense, employees are exposed to digital devices that enable invasive surveillance, affecting their privacy and generating risks of corporate surveillance, which in turn creates vulnerability to personal data (Goñi, 2019, pp. 63-64). Similarly, discrimination in candidate selection becomes a risk, as it can reinforce hiring bias if algorithms are not well designed, undermining equity in access to employment (Cheatham et al., 2019).

AI challenges in the work context: constant training, which requires an AI-driven transformation of the labor market, and requires workers to adopt a lifelong learning approach. Adapting to an automated environment requires acquiring advanced digital skills, such as creative thinking and emotional intelligence, which are key elements for collaborating with emerging technologies (Corvalán, 2019, pp. 43-44). Education and professional development must adapt to the new demands of employment, promoting training programs that guarantee the effective insertion of workers into the digital age.

Legislation and regulation must also be implemented to mitigate the adverse effects of automation, primarily by establishing a robust legal framework that protects workers’ rights and ensures the ethical use of AI. The European Union has made progress in drafting legislative proposals that address the impact of AI on employment, promoting regulations that guarantee transparency and fairness in their implementation (Parra and Concha, 2021, p. 9). These regulations are crucial for preventing discriminatory practices and establishing the boundaries of automating work processes.

Regarding the humanization of AI, “co-boosting” refers to the collaboration between humans and machines, representing an alternative approach to integrating AI without completely replacing human work. In this model, technology acts as a complement, enhancing workers’ capabilities and enabling greater intelligence in the management of work tasks (Corvalán, 2019, pp. 44-45). This balanced interaction could optimize productivity and improve job quality without generating a total replacement of workers by automated systems.

In terms of privacy, the use of digital tools in the workplace has raised concerns about employee privacy.

Massive data collection, constant monitoring, and digital surveillance can affect individual rights, requiring protection mechanisms that regulate the management of personal information within companies (Goñi, 2019, pp. 63-64).

To avoid violations, organizations must ensure the responsible use of AI systems, guaranteeing that workers’ privacy and well-being are protected.

As a final major challenge, the idea of reconfiguring the labor market must be considered, because while AI may replace certain professions, it is also behind the creation of new opportunities in modalities such as freelance work and the collaborative economy. Digitization is enabling the development of more flexible forms of employment, where workers can adapt their skills to emerging sectors and participate in dynamic projects without depending exclusively on traditional employers (Corvalán, 2019, pp. 42-43). This transformation requires continuous updating of skills to capitalize on the benefits of a digitized work environment.

References

• Arrigain García, I. (2020). Regulación de las invenciones relacionadas con Inteligencia Artificial en el campo de la Propiedad Intelectual. Universidad Pontifica Comillas. 
• Bravo, J., García, A. y Schlechter, H. (2018). Automatización e inteligencia artificial: desafíos del mercado laboral. Centro Latinoamericano de Políticas Económicas y Sociales.
• Cheatham, B., Javanmardian, K. y Samandari, H. (2019, 26 de abril). Enfrentando los riesgos de la inteligencia artificial. McKinsey & Company. https://www.mckinsey.com/capabilities/ quantumblack/our-insights/confronting-the-risks-of-artificial intelligence/es-CL#/
• Corvalán, J. G. (2019). El impacto de la inteligencia artificial en el trabajo. Revista de Direito Econômico e Socioambiental, 10(1), 35 51. https://doi.org/10.7213/rev.dir.econ.soc.v10i1.25870
• Coeckelbergh, Mark. (2020). AI ethics. Hoe omgaan met artificiële intelligentie? 
• González, N (2024). Algunos parámetros jurídicos y éticos en torno a la inteligencia artificial. Una visión humanista de la tecnología. Editoral ABC
• Goñi, J. L. (2019). Innovaciones tecnológicas, inteligencia artificial y derechos humanos en el trabajo. Documentación Laboral, 2(117), 57-72. https://dialnet.unirioja.es/servlet/ articulo?codigo=7095888
• Massachusetts Institute of Technology. (n.d.). The AI Risk Repository. Retrieved April 26, 2025, from https://airisk.mit.edu/
• Parra, D. y Concha, R. (2021). Inteligencia artificial y derecho. Problemas, desafíos y oportunidades. Vniversitas, 70, 1-25. https://doi.org/10.11144/Javeriana.vj70.iadp

AI law: different rules for different risk levels

The new regulation sets out obligations for suppliers and users based on the risk classification level of the AI. Whilst many AI systems pose only a minimal risk, all must be assessed.

Unacceptable risk

AI systems that present an unacceptable risk and are banned in the EU are as follows:

• Cognitive manipulation of the behavior of specific vulnerable individuals or groups: for example, voice-activated toys that encourage dangerous behavior in children;
• Social rating: classifying individuals based on their behavior, socio-economic status, or personal characteristics;
• Real-time and remote biometric identification systems, such as facial recognition in public spaces.

There are a few exceptions to this qualification. For example, “a posteriori” remote biometric identification systems, where identification occurs after a significant delay, will be authorized for the prosecution of serious crimes, provided there is prior judicial approval.

High Risk

AI systems that undermine security or fundamental rights are considered high-risk and are divided into two categories.

1. AI systems are used in products subject to European product safety legislation. These include toys, aviation, automobiles, medical devices and elevators.
2. AI systems in eight specific areas must be registered in an EU database:
   • Biometric identification and categorization of natural persons;
   • Management and operation of critical infrastructures;
   • Education and vocational training;
   • Employment, management of workers, and access to self-employment;
   • Access to and enjoyment of essential private services, public services, and benefits;
   • Management of migration, asylum, and border controls;
   • Assistance in interpreting and applying the law.

All high-risk AI systems will be assessed before they are put on the market and throughout their lifecycle. Citizens will have the right to lodge complaints about AI systems with specific national authorities.

Transparency requirements

Generative AI, such as ChatGPT, will not be considered high risk, but will have to comply with transparency requirements and EU copyright legislation:

• Reveal that AI has generated content;
• Design the model to prevent it from generating illegal content;
• Publish summaries of copyrighted data used for training;

General-purpose AI models with a high impact and the potential to present systemic risk, such as the GPT-4 advanced AI model, will be required to undergo in-depth assessments and report any serious incidents to the Commission.

For a human management of Artificial Intelligence - Introduction through ethics and governance

(1) For a human management of Artificial Intelligence – Introduction through ethics and governance (Part 1)

(2) For a human management of Artificial Intelligence – Introduction through ethics and governance (Part 2)

(3) For a human management of Artificial Intelligence – Introduction through ethics and governance (Part 3)

(4) For a human management of Artificial Intelligence – Introduction through ethics and governance (Part 4)

(5) For a human management of Artificial Intelligence – Introduction through ethics and governance (Part 5)

(6) For a human management of Artificial Intelligence – Introduction through ethics and governance (Part 6)

(7) For a human management of Artificial Intelligence – Introduction through ethics and governance (Part 7)