Maya, Ben and Rosa each look at one tool in their week that sorts, scores or chooses people.
Maya thinks about the photo app. Ben thinks about the reading app and the online test system. Rosa thinks about a tool that offers to sort job applications.
Each of them asks the two questions. Who is missing? Who is misjudged?
Read your own track first. The idea is the same in each: a tool can serve some people less well, so a person watches for it.
A photo tool that mixes up some family members is a small problem. But the same pattern can show up in tools that matter more.
AI4K12 warns that biased training data can lead to some people being less well served than others.
You can talk about this with your kids. If a tool gets someone wrong, that is a flaw in the tool, not in the person.
The Department of Education report gives two examples of algorithmic bias.
One is a voice recognition system that does not work as well with regional dialects.
The other is an exam monitoring system that may unfairly flag some groups of students.
A flag is a mark for attention, not proof of anything. The same report says teachers keep the power to decide what to do.
So a teacher looks at every flag and listens to every student before acting on a score.
Suppose a tool offers to screen job applicants or choose who sees an ad. It learns from data, and that data may leave people out.
If past data left some customers or applicants out, a tool may keep leaving them out. Ask who it skips.
The SBA says to have another person review AI output, and to make sure a person checks every AI message and campaign.
For any tool that sorts people, that review should include the two questions: who is missing, and who is misjudged?
| Statement | True or false? |
|---|---|
| The Department of Education names exam monitoring that may unfairly flag some groups. | ? |
| A flag from a tool proves a student did something wrong. | ? |
| A tool that learns from past data can repeat who was left out before. | ? |
| The SBA says a person should check every message AI generates. | ? |
Sources: AI4K12, Five Big Ideas in Artificial Intelligence (poster, version 2) · US Department of Education, AI and the Future of Teaching and Learning (2023) · US Small Business Administration, AI for small businesses
Tomorrow: a short review of the week.