# Seeing Beyond the Algorithm | Jenna Hammoud | TEDxYouth@JeffersonStreet Source: https://www.youtube.com/watch?v=bjT3tmiRGCc Recap page: https://rapidrecap.app/video/bjT3tmiRGCc Generated: 2025-12-03T17:54:42.15+00:00 --- ## Quick Overview Jenna Hammoud argues that Artificial Intelligence (AI) systems, trained on historically biased data, perpetuate and amplify societal inequalities, demonstrated by her experiment where an AI voice recognition system failed to recognize her voice due to bias favoring higher-pitched voices, leading her to call for greater AI literacy and ethical development to ensure future technology is inclusive and human-centered. **Key Points:** - The speaker conducted an experiment by prompting ChatGPT to generate images of a doctor and a teacher, consistently receiving images depicting men, illustrating inherent gender bias in AI training data. - The speaker's personal experience involved an AI voice recognition system failing to recognize her voice, which she later learned was not malfunctioning but was trained predominantly on higher-pitched female voices, leading to the exclusion of her voice pitch. - A major tech company developed an AI hiring tool trained on ten years of resumes submitted to the company, which predominantly favored male candidates, resulting in the system discriminating against women. - The speaker cites Joy Buolamwini's book, "Invisible Women," which exposes how voice recognition systems often fail to accurately detect darker-skinned female faces or recognize women's voices. - The speaker concludes that AI tools are not neutral; they reflect the biases present in their training data, which often mirrors historical societal inequalities. - The solution proposed is to challenge these systems through increased AI literacy and demanding ethical, inclusive development from major technology companies. - The speaker explicitly states that the goal is to build a future that is inclusive, fair, and human-centered, rather than letting AI reinforce existing biases. ![Screenshot at 00:48: The speaker shows the scale of the problem by stating that 2.5 billion prompts are processed by ChatGPT daily, emphasizing the massive, daily reliance on AI that mirrors societal biases.](https://ss.rapidrecap.app/screens/bjT3tmiRGCc/00-00-48.png) **Context:** Jenna Hammoud delivers a TEDxYouth@JeffersonStreet talk focused on the pervasive issue of bias embedded within Artificial Intelligence (AI) systems. She uses personal anecdotes and external research, such as the work of Joy Buolamwini, to illustrate how AI, trained on historical data reflecting societal prejudices, often perpetuates and even amplifies discrimination against women and minorities in areas like hiring and voice recognition. ## Detailed Analysis Jenna Hammoud argues that the widespread trust in Artificial Intelligence (AI) for making billions of daily decisions is dangerous because these systems inherit and reinforce existing human biases found in their training data. She illustrates this using two main examples: first, prompting ChatGPT to generate images of doctors and teachers resulted in images almost exclusively depicting men, proving gender bias in visual generation. Second, she recounts her personal failure to be recognized by a voice recognition system, which turned out to be biased against her vocal pitch because it was trained on data favoring higher-pitched female voices. She references Joy Buolamwini’s book, "Invisible Women," which documented how commercial facial recognition systems failed to accurately detect darker-skinned female faces. Hammoud points out that a major tech company's AI hiring tool, trained on ten years of resumes, systematically favored male applicants because the historical data reflected that imbalance. She asserts that these tools are not neutral; they are reflections of our societal inequalities. To combat this, Hammoud calls for collective action: questioning AI responses, increasing AI literacy, and challenging large tech companies to develop ethical, inclusive AI systems that are representative of everyone, not just the privileged groups reflected in historical data. ### Introduction to AI Bias - Audience interaction regarding ChatGPT use - Stating 2.5 billion daily prompts - AI inheriting biases from data - Example of AI generating only male doctors/teachers ### Personal Anecdote - Speaker tested an AI voice recognition system that failed to recognize her voice - The system was trained to recognize higher-pitched voices, excluding her pitch ### Research and Examples - Citing Joy Buolamwini's research on facial recognition failing darker-skinned women - Mentioning a tech company's AI hiring tool favoring men due to biased historical training data (10 years of resumes) ### The Core Problem - AI systems are not neutral; they mirror and reinforce human biases and historical inequalities, such as in voice recognition and hiring selection ### Call to Action - Urging the audience to question AI responses, increase AI literacy, and challenge tech companies to build inclusive, human-centered technology that avoids reverting to male norms. ![Screenshot at 00:01: The title slide for the event: TEDxYouth@JeffersonStreet, defining 'x' as an independently organized TED event.](https://ss.rapidrecap.app/screens/bjT3tmiRGCc/00-00-01.png) ![Screenshot at 00:11: A visual montage transition showing the word 'Ideas' over a background of question marks, leading into the next segment.](https://ss.rapidrecap.app/screens/bjT3tmiRGCc/00-00-11.png) ![Screenshot at 00:24: A heavily stylized, red-tinted graphic displaying the text: 'STEAM with a twist,' referencing the event's theme.](https://ss.rapidrecap.app/screens/bjT3tmiRGCc/00-00-24.png) ![Screenshot at 01:40: The speaker prompts the AI to generate an image of a doctor, setting up the demonstration of bias.](https://ss.rapidrecap.app/screens/bjT3tmiRGCc/00-01-40.png) ![Screenshot at 03:00: The speaker delivers the main thesis, asking the audience why they trust algorithms so much when they are not neutral.](https://ss.rapidrecap.app/screens/bjT3tmiRGCc/00-03-00.png)