A company develops a program to help high school counselors suggest career paths to students. The program uses a machine learning algorithm that is trained on data from social media sites, looking at a user's current job title, their background, and their interests. After counselors use the program for a while, they observe that the program suggests jobs in STEM (science, technology, engineering, math) much more often for male students than for other students. What is the most probable explanation for the bias in the career suggestions?