To me, a linear regression is part of Machine Learning, and, thus, part of the broader vision of AI. Even though linear regressions are old statistical models and probably existed long before the term ML.The linear regression algorithm is learning from data (i.e. improves the line fitting after observing more data. hence, it is ML in my book) -- it just has a very limited hypothesis space. It will only ever fit a straight line (or hyperplane, in the general case). It is not a general learner like a Deep Neural Network which can approximate any function.