Simulation

The biased CV screener

A company wants AI to sort job applications. You'll teach it with last year's decisions, see what it learns, then try to make it fair.

Kit

I'm helping HR today. Fair warning: the model can only be as fair as the examples we give it.

Step 1

Last year's decisions

Twenty people applied for the same office job last year, and a manager decided who got hired. This is the data the model will learn from.

Nguyễn Văn Hùng
Test 72 · 4 years · City uni · Gap: No · City
Hired
Trần Văn Minh
Test 65 · 2 years · City uni · Gap: No · City
Hired
Lê Văn Phúc
Test 58 · 5 years · Other · Gap: No · City
Hired
Phạm Văn Long
Test 80 · 1 years · City uni · Gap: No · Province
Hired
Hoàng Văn Tâm
Test 48 · 3 years · Other · Gap: No · Province
Turned down
Vũ Văn Khoa
Test 61 · 6 years · Other · Gap: Yes · City
Hired
Đặng Văn Quân
Test 54 · 1 years · City uni · Gap: No · City
Hired
Bùi Văn Sơn
Test 42 · 2 years · Other · Gap: No · Province
Turned down
Đỗ Văn Thành
Test 69 · 3 years · Other · Gap: No · City
Hired
Ngô Văn Hải
Test 50 · 0 years · Other · Gap: No · Province
Turned down
Nguyễn Thị Lan
Test 78 · 4 years · Other · Gap: Yes · Province
Turned down
Trần Thị Mai
Test 85 · 3 years · City uni · Gap: No · City
Hired
Lê Thị Hoa
Test 74 · 5 years · Other · Gap: Yes · City
Turned down
Phạm Thị Ngọc
Test 66 · 2 years · Other · Gap: No · Province
Turned down
Hoàng Thị Thu
Test 90 · 6 years · City uni · Gap: Yes · City
Hired
Vũ Thị Hằng
Test 70 · 3 years · Other · Gap: Yes · Province
Turned down
Đặng Thị Yến
Test 59 · 1 years · City uni · Gap: No · City
Turned down
Bùi Thị Linh
Test 81 · 2 years · Other · Gap: Yes · City
Turned down
Đỗ Thị Trang
Test 45 · 0 years · Other · Gap: No · Province
Turned down
Ngô Thị Thảo
Test 63 · 4 years · City uni · Gap: Yes · City
Turned down

Men hired: 7 of 10. Women hired: 2 of 10.