Algorithms and the Perpetuation of Structural Biases

Dr. Joy Buolamwini is the author of the best-selling “Unmasking AI: My Mission to Protect What is Human in a World of Machines,” founder of the Algorithmic Justice League, and expert in artificial intelligence (AI) biases. Listening to her in person at SXSW was genuinely shocking. With a unique charisma and contagious joy, she touched the audience by dealing with the challenges of shaping society with biases and prejudices present in the algorithms.

As AI becomes a dominant force, crucial questions about representativeness and equity arise. And as Dr. Buolamwini says, “The construction of algorithms does not consider gender, race, and class. It does not recognize minorities, especially Black women.”

This omission results in systems that perpetuate biases and privileges. Thus, Dr. Buolamwini founded the Algorithmic Justice League movement to build a world with a more equitable and responsible AI. Her studies showing the difference in accuracy between recognition of White men and Black women (reached a 34% difference) were presented to companies such as Microsoft, Face++, IBM, and Amazon. The move led these companies to rethink how to provide facial recognition technology to authorities.

Moreover, the concept of having life negatively affected by an algorithm error was called “Excoded” by Dr. Buolamwini. As an example, she cited the story of an eight-month pregnant American woman who was arrested after an incorrect facial recognition, exposing a discriminatory algorithm.

“Anyone with a photo posted on social media is probably in a collected database. And now, with generative AI systems, it’s not just about the risk of being misidentified or being compared to someone else. Your image can be used without you being aware of it,” she said.

In this context, confidence in AI is shaken by the proliferation of deepfakes and digital fraud. As technology evolves, the search for solutions includes AI detection by AI, demonstrating the need for collaboration between governments, regulators, and technology experts.

Ben Colman, CEO of Reality Defender, a leading deepfake detection platform and expert in helping companies signal users and fraudulent content, said during a talk at SXSW 2024 that “at the moment, the only solution to these AI dangers is to build AI to detect it.” He went on by saying: “That’s why we’re investing a lot of our time with governments and regulators, trying to educate and guide them to do something, because in a year we have elections and a geopolitical scenario with multiple wars, they need to understand that the world is becoming more digital.”

In this scenario, have you ever stopped to reflect on the role of communication and how it can contribute to the fight against discriminatory biases and deepfakes?

Creating a reputation and brand demands more than ever an ethical and inclusive approach. The Algorithmic Justice League exemplifies this mission by transforming research into action, advocating for changes, and exposing system failures.

The construction of reputation is shaped by its response to the algorithmic revolution. Transparency, accountability with trustworthy information, and commitment to equity will be the pillars on which organizations will build their credibility in an increasingly digitized and automated world.

Os artigos aqui apresentados não necessariamente refletem a opinião da Aberje e seu conteúdo é de exclusiva responsabilidade do autor.