Ethics in Artificial Intelligence Systems

There is not a day in which we do not see some news related to Artificial Intelligence (AI) and, although there is a common position regarding the benefits that it can generate in different areas such as health, education, environment, etc., the development of AI-based systems generates certain ethical challenges that can result in wide-ranging risks, since they will be used worldwide.

We could ask ourselves, how can a technology that should be designed to facilitate work, decision-making and contribute to the improvement of people’s lives, have a negative impact if it is not designed and monitored properly?

Taking as a reference the reflections of Coeckelbergh (AI Ethics, 2021): “AI will progressively increase its capacity for intentional agency, replicating and replacing human agency, generating the problem of the absence or dissolution of ethical responsibility in technological systems”.

In sum, some of the main ethical risks we face with AI Systems (AIS) would be:

  • Increase in the capacity of AIS to the detriment of human decision-making capacity.
  • Existence of discriminatory biases in algorithms, influencing decision-making at all levels and losing control over the effects of AIS.
  • Encouraging opacity and restricting accountability as AIS become more complex.
  • Potential to exacerbate social and/or economic inequalities.
  • Increased carbon emissions from disproportionate energy use.

Therefore, AIS should be used to improve decision making, but never replacing or vitiating the will of human intelligence, since the worst scenario in which we could move would be one in which the human being renounces, either voluntarily or forcibly, to make binding decisions on a daily basis.

According to Luciano Floridi, the basic principles of bioethics would be applied to AIS: beneficence, non-maleficence, autonomy and justice, in addition to explainability, which would imply intelligibility and transparency of AIS, and accountability.

UNESCO’s ethical recommendations focus on the following principles (UNESCO, 2022):

  • Proportionality and harmlessness
  • Equity and non-discrimination
  • Safety and security
  • Sustainability
  • Privacy and data protection
  • Oversight and human decision making
  • Transparency
  • Responsibility and accountability
  • Adaptive stakeholder governance and collaboration
  • Awareness and education

These principles are homologous to those contained in the Montreal Declaration and emphasize the need to safeguard human decisions and responsibilities, as well as the importance of protecting personal data, transparency and accountability of companies and states.

Once the risks and ethical principles applicable to AIS have been identified at a high level, the following UNESCO recommendation would be plausible: “should people have good reasons to trust that AI systems can bring individual and shared benefits, while taking appropriate measures to mitigate risks?” (UNESCO, 2022).

Let us hope that these ethical principles do not remain merely a wish list, and that the economic and political context in which we are living does not allow us to be able to make the most of these ethical principles, but that they are not just a wish list.

See also in:

Speak Your Mind