9. Bias is baked into the patterns.
Chatbots learn from text written by people. People have biases. The chatbot absorbs them along with everything else.
This can show up quietly — defaulting to "he" for a doctor, giving more detailed answers about some cultures than others, or reflecting stereotypes without flagging them. The companies that build chatbots work to reduce this, but it cannot be fully eliminated.
What you can do: Notice when a response seems one-sided or makes assumptions. Ask the chatbot to consider a different angle. You are the critical thinker in the conversation — the chatbot is not.
Lesson 9 — Bias is baked into the patterns
Written by people
Billions of pages of text — including human biases, assumptions, and gaps.
Chatbot learns patterns
It absorbs everything — useful patterns and biased ones alike.
Bias can appear in responses
Defaults, assumptions, and gaps you may not notice at first glance.
You are the filter
Notice assumptions. Ask for other perspectives. Apply your own judgment.
The chatbot can't catch its own blind spots. You can.
More info: Bias in AI — What It Looks Like and What You Can Do