Can Artificial Intelligence Hear Us?
Artificial Intelligence (AI) has the potential to improve our lives and foster innovation across various fields. However, a concerning pattern is emerging: the reinforcement of unconscious biases against women in AI systems.
Biases influence our decisions and actions, often without our awareness. AI algorithms and models, created and trained by humans, inherit the conscious and unconscious biases of their developers. This is evident in facial recognition technology, which is more accurate at identifying male faces than female ones, especially women with darker skin tones, leading to disproportionate discrimination in security-related areas.
In language processing and text generation, AI models also reinforce stereotypical gender roles. Chatbots and voice assistants often have female voices and traits like submissiveness, perpetuating social prejudices.
The illustration for the article was created with the help of ChatGPT-4 on December 23, 2023.
Prompt: “Please generate an image about unconscious biases against women that appear or are even amplified when using artificial intelligence.”
One root cause of these biases is the training data. When algorithms are trained on historically biased data, they reproduce and even amplify these biases. Texts from past decades that contain gender stereotypes are used by AI to generate new content.
Addressing this issue requires identifying and correcting biases in the training data. This involves diverse development teams and critical reflection on data collection and usage. Creating transparent AI models that explain their decision-making processes can also expose and correct potential biases.
Governments and organizations must develop policies and standards that integrate ethics and diversity into AI development. The “European Artificial Intelligence (AI) Act” is a pioneering initiative addressing this issue.
As Soroptimists, we can significantly influence responsible AI development by raising awareness and demanding adherence to ethical principles and data protection standards. We must question AI outcomes and their impacts, taking responsibility for the rules and algorithms guiding AI decisions.
Each of us has a role in this transformation. By providing feedback on AI results, we contribute to improving algorithms. We must act now to demand ethical principles in AI development. Soroptimists are a global voice for women – let AI hear us loud and clear.
Reference
This article was created with the help of ChatGPT. The proposed text was edited by the author, based on her lectures from 2021, 2022, and 2023.