Nathan,

Thanks for sharing your paper, but the first two sentences of the
abstract contain some false claims that undercut AI's long and
successful history.

> Artificial intelligence (AI), which enables machines to learn to
> perform a task by training on diverse datasets, is one of the most
> revolutionary developments in scientific history.

AI is much broader than just machine learning and the first serious
applications occurred in the late 1970s and early 1980s in the form
of hand-crafted expert systems.

> Although AI, and especially deep learning, is relatively new, it has
> already had a transformative impact on medicine, biology,
> transportation, entertainment, and beyond.

Machine learning is far broader than deep learning. The field initially
grew out of expert systems and aimed to automate their construction.
By the mid-1990's, it had produced many fielded applications:

Langley, P., & Simon, H. A. (1995). Applications of machine learning and
    rule induction. Communications of the ACM, 38, November, 55-64.
    http://www.isle.org/~langley/papers/mlapps.cacm95.pdf

This article reviews successes of decision-tree and rule induction, but
other papers at the same reported similar results with neural networks,
case-based approaches, and other paradigms.

> To this end, a critical step is to ensure an AI-ready workforce through
> education.

I agree that we should be training people in AI, but their education
should cover other mature technologies, too, like automated planning
and answer set programming, which do not rely on machine learning.

Best -Pat