AI and ethics: the balance between progress and responsibility

More than just technology: why AI is also an ethical issue

Artificial intelligence (AI) is revolutionizing the world of work in a wide range of industries and significantly improving efficiency. The publication of ChatGPT by OpenAI in 2022 marked the decisive turning point for the technology. Its capabilities, which range from text generation to coding, open up numerous possibilities and make AI applicable in almost all areas.

This also applies to the financial sector. Here in particular, the technology can play an important role in long-term efficiency, security and customer focus and can help with risk management, customer interaction or the automation of processes, for example. However, with the increasing performance, availability and usability of AI also come risks that need to be considered.

Man-made systems bring with them problems ranging from bias and the generation of incorrect content to societal impacts such as job losses. This can have devastating effects on both human lives and businesses.

This example clearly demonstrates that the responsible use of AI in compliance with ethical standards, a moral compass and a deep understanding of the technology is increasingly crucial. The lack of attention to this area can not only burden customers, but also companies with reputational and legal risks as well as high financial penalties. To stay on top of the situation, companies should therefore start incorporating ethical issues into their discourse and strategies at an early stage.

Black box, bias and responsibility: the ethical construction sites of AI

The potential problems and ethical concerns associated with artificial intelligence relate to various interconnected areas.

Some key topics such as a lack of transparency and explainability, bias and discrimination, security risks, social inequality, responsibility and liability are therefore examined in more detail below.

Transparency and explainability: AI models are so-called “black boxes”. It is often unclear exactly how decisions are made, which makes it difficult to understand and trust the systems.

Bias and discrimination: AI systems can adopt and reinforce existing prejudices from the training data. This leads to discriminatory decisions, for example in application procedures or the granting of loans.

Security risks: Artificial intelligence can be used to create deepfakes or for the automated dissemination of fake news, which can jeopardize democratic processes and be used for manipulative or criminal purposes. If AI systems are given too much decision-making power, human autonomy can also be undermined.

Social inequality: Automation through artificial intelligence can replace many jobs. Equally, unequal access to artificial intelligence can exacerbate social injustice.

Responsibility and liability: If an AI makes a faulty decision, it is often unclear who is responsible: the developer, the operator or the system itself?

These challenges make it clear that the responsible use of artificial intelligence is not only a technical task, but above all an ethical and social one that requires clear guidelines, transparency and interdisciplinary cooperation.

Basic principles of AI ethics and their regulatory implementation

AI ethics is a multidisciplinary field. It examines principles that help distinguish right from wrong and promote positive impacts of artificial intelligence while reducing risks and negative consequences.

How artificial intelligence works depends largely on how it is designed, programmed, trained, adapted and used. The ethical debate on AI aims to establish a framework of ethical principles and guidelines that takes into account all phases in the life cycle of an AI system.

Companies, governments and researchers have already begun to develop various frameworks for dealing with artificial intelligence in order to address the concerns surrounding artificial intelligence. The seven ethical principles of the EU Commission/OCED/EU High-Level Expert Group on AI are often cited as the basis for this, which are outlined below:

• Human action and supervision have priority: AI systems should support people, not replace them. Fundamental rights must not be violated.
• Technical robustness and safety: Artificial intelligence must be safe, reliable and resilient. Accountability: It must be clear who is responsible for the actions and decisions of an AI system.
• Diversity, non-discrimination and fairness: Artificial intelligence must not put anyone at a disadvantage. Fairness in development and application is essential.
• Transparency: AI decisions should be comprehensible and explainable.
• Social and ecological well-being: Artificial intelligence should serve the common good, promote social and ecological sustainability and not just follow economic interests.
• Privacy and data protection: The protection of personal data and respect for privacy must be guaranteed in the development and use of artificial intelligence.

The EU AI Act, which came into force in 2024, also takes up the aforementioned ethical principles and translates them into legally binding regulations. By classifying AI systems into four risk categories (see illustration), safety, compliance with human rights and social welfare are ensured. The transparency and documentation requirements issued for certain systems ensure traceability and review. In addition, the AI Act refers to the GDPR and requires AI systems to be designed and operated in compliance with data protection regulations.

Conclusion: Governance as the key to trustworthy artificial intelligence

In order to introduce and/or operate artificial intelligence in line with the ethical principles and values of an organization and the expectations of stakeholders, a governance program is essential.

This enables the AI lifecycle to be monitored through internal guidelines and processes. To this end, clear roles and responsibilities should be defined for all persons involved.

At the same time, all stakeholders along the entire AI lifecycle should be sensitized and trained for responsible development and use. In order to ensure the structured and safe use of artificial intelligence, processes should be established that regulate the development, management, monitoring and communication of AI systems and their potential risks. In addition, suitable tools can be used that aim to continuously improve the performance and trustworthiness of artificial intelligence throughout its entire life cycle.

In summary, this means If companies see artificial intelligence as a supporting tool that is based on ethical principles and focuses on people, its use can be made sustainable, secure and trustworthy.

A holistic governance program forms the foundation for a responsible and future-oriented AI strategy.