AI for Solopreneurs: The AI Hat Podcast

The Hidden Biases in Your AI (And How to Fix Them) with Dr. Nici Sweaney

13 snips
Sep 2, 2025
Guest
Nici Sweaney
Dr. Nici Sweaney is a globally recognized leader in ethical AI and founder of AI Her Way, helping organizations implement fair AI solutions. She dives deep into how hidden biases in AI can silently impact customer relationships and bottom lines. Dr. Sweaney emphasizes the importance of auditing AI systems and shares strategies to confront gender bias. She advocates for a collective responsibility in AI equity, highlighting the potential for businesses to harness ethical AI for success, turning challenges into transformative opportunities.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

LLMs Reinforce Dataset Stereotypes

  • Large language models predict patterns and often reinforce stereotypes based on their training data.
  • That default pattern-seeking causes gendered outputs unless explicitly corrected.
ANECDOTE

Male Bears From An Image Prompt

  • Mike asked an image generator for talking bears and got almost all male bears.
  • He recognized the gender imbalance immediately because of his prior focus on elevating female representation.
ADVICE

Prompt Models To Check For Bias

  • In prompts, ask models to remove identifying demographic information before making value judgments.
  • Also ask for alternate perspectives and counterarguments to surface hidden bias.
Get the Snipd Podcast app to discover more snips from this episode
Get the app