Notes   /   22 February 2019

Understanding and Analyzing Image Search Results

This is a group activity intended to help you get to know your teammates and learna bit about how algorithms like the one that drives Google's Image Search construct, influence, or respond to social norms. In this activity, you will compare search results for different terms.

Step 1: Choose some Terms of Comparison

Choose a pair of related or similar careers or identity categories like "doctor","nurse","professor","teacher", "parent","grandparent","criminal", "baby", "entrepreneur", "technician", "programmer", etc.

For each search result, run a Google Image search, then record evident demographic characteristics in each. Make a table listing categories like gender, race, age, or any other representational characteristic.

Step 2: Discuss the results

• What are the most consistent characteristics in your search results?
• How strong is the association between those characteristics and those careers?
• How does the result change if you're logged in to your Google account?
• How do the results in DuckDuckGo or Bing compare to Google's?
• Are these results and characteristics reinforcing stereotypes or responding to them?
• What makes google associate an image with a keyword in the first place?

Step 3: Share the result of your discussion with the class