Before AI Learns Bias: How Historical Inequity Shows Up in Data | Angel Ceballos

Before an AI model makes a single decision, it is trained on decades of historical data. Data that often reflects human bias, systemic inequity, and structural blind spots. In this powerful talk from the 1st Annual AI for Women Conference, Angel Ceballos breaks down how historical inequity gets coded into modern algorithms and what we must do to build fair, responsible AI systems. Recorded live at American Underground in Durham, NC, and hosted by the Center for Practical AI. KEY TAKEAWAYS & HIGHLIGHTS: • How historical and societal biases silently enter training datasets • The "garbage in, garbage out" trap in machine learning models • Practical frameworks for auditing data for equity before model deployment • Why diverse perspective in tech leadership is essential for responsible AI FEATURED SPEAKER: • Angel Ceballos — CEO, SeraphicGuardian & Fractional CDO ORGANIZED BY: Center for Practical AI (https://www.cp-ai.org/) Location: American Underground (Durham, North Carolina) #WomenInAI #ArtificialIntelligence #AIBias #EthicalAI #DataEquity #TechLeadership #CenterForPracticalAI #DurhamNC #WomenInTech #ResponsibleAI