Over the last few decades, technology has become increasingly intelligent. Technology is no longer a passive tool that supports a single human in their work, but an active teammate that collaborates and learns as a critical entity of the team.  To date, human-machine (HM) teaming research has primarily focused on the machines – how to design them, what their capabilities are, and how they can “learn.” This conceptual paper takes the opposite view, focusing on the importance of selecting and training humans to be effective HM teammates. To that end, this paper will address two questions: What unique skills do humans need to work well with machines as teammates, and how are those skills different from those required for effective human-human interactions? The challenges that HM teams face drive the identification of the human skills. For example, humans are fundamentally biased to anthropomorphize machines and expect them to act like other humans (Proudfoot, 2011). Consequently, humans expect to understand and predict how and why machines are making their decisions. When machines do not act in accordance with human expectations, trust and coordination between humans and machines quickly break down (Mueller, Hoffman, Clancey, Emrey, & Klein, 2019).  To mitigate this effect, we build machines with explainable AI to provide humans with insight into their decision making (Mueller et al., 2019). We can also improve HM teaming by selecting humans who have individual traits such as openness to new experiences, tolerance for ambiguity, and high propensity to trust. Humans can be trained on perspective taking skills to understand how machines make decisions (Galinsky, Ku, & Wang, 2005). In addition, identifying the skills humans need to work with machines, this paper will make suggestions for how to train humans and machines together for effective HM team performance (Nikolaidis & Shah, 2013).