Fixing Students’ Early Math Skills is Critical for the Future of ML and all STEM

Namrata Vaswani, Mohamed Y. Selim, Renee Gibert, Aditya Ramamoorthy

Keywords: K-12, school, math skills, ML, success, equity

TL;DR: Improving early math skills of all students (and especially of the most vulnerable) students is critical for the success of ML and all STEM

Abstract:

This position paper argues that fixing early math skills of students is critical for the future of machine learning (ML) and all science, technology, engineering or mathematics (STEM) professions. It also describes some ways ML and STEM scientists can help.

A common requirement for becoming a successful ML or STEM scientist or professional is fluency in primary and early secondary school mathematics (arithmetic and basic algebra in particular). Math learning is cumulative and thus, these early skills are critical both for learning high school or college math, physical sciences and computing courses well, as well as for learning to correctly write code for ML or scientific applications. Because of various reasons, including circumstances beyond their control, many students are not able to build a strong foundation in arithmetic in elementary school. This leads them to struggle with pre-algebra skills and then, with (scalar) algebra and everything after that. What can ML and STEM scientists and educators do to remove or to reduce the effect of some of the learning barriers? We describe two partial solutions. The first is running, or participating in, out-of-school math support programs (here ``support'' can mean tutoring sessions, math practice sessions, or just sharing free or low-cost math learning resources and encouraging practice) to help close the early learning gap between those with such math awareness at home and those without. The second is that ML or STEM educators and researchers could and should step in and comment on the long-term impact of modern elementary and middle school math education policies (such as very little math practice, almost no homework, either no testing or not informing students and parents about testing, etc). Since it is practically difficult to conduct rigorous long-term education research, most current policies are based on short-term (2-3 years or less) research

Submission Number: 406