Machine Learning is becoming ubiquitous in the realms of science, business, law enforcement, and even the arts.  Many people don’t realize how rooted in “physics thinking” the foundations of machine learning algorithms are, but physics teachers are well positioned to teach students to use these tools and how to understand their limitations.  I have developed a three-week general education course on machine learning for non-scientists, and I adapted several activities for an introductory laboratory sequence in a course on Galaxies and Cosmology.  I will report on the successes and challenges of these endeavors and make recommendations on how students can be prepared at multiple educational levels to engage with the opportunities and dangers these tools represent.

The Department of Physics and Astronomy at Eastern Michigan University requires students who have chosen a physics major or minor to take a course introducing them to scientific computation.  Because that course has evolved from introducing both spreadsheets and Python to focus solely on Python, there was an opportunity and need to incorporate computational modeling with spreadsheets in the laboratory portion of the calculus-based introductory mechanics course.  Students now learn the Leapfrog method of numerical integration to computationally model falling with air resistance, the spin-down of a rotary motion sensor and magnetically damped fidget spinner, and the oscillations of spring-mass systems and physical pendula.  We describe the content and sequencing of these labs and how these new labs are consistent with an increasing focus on student skill development.

Interestingly enough, if you stop and think about it, science makes no sense from a observational perspective. Consider F = ma, what does the “=“ symbol reference?  Or for that matter what does it mean to multiply mass with acceleration? K.E. = 1/2 mv^2, raises more questions: what does v^2 look like; or why does K.E. only equal half of mv^2?  If equations “described" exactly what is taking place not experimentally but experientially, learning would be vastly easier.  In other words, equations can be expressed in a form where the are the means of describing “this” in terms of “that” which requires a “relationship" between the “this” and “that”. Expressed in this way, the laws of nature take on a completely different meanings.