The possibility of human-AI (HAI) collaboration comes from creating groups that perform better than either humans or AI could on their own. However, research shows there is a big difference between this ideal teamwork and how it actually works in practice. To understand this gap, we look at how the field has developed its research methods over time. We see a clear path from early studies that measured the difficulty of working together, to building ideas about how humans and AI can interact, and finally to testing these systems in both controlled experiments and real-life situations. Our review shows the field has grown from being focused only on how well AI algorithms work, to a better understanding of human-centered aspects like trust, independence, and detailed control. By looking at this progress, we find important areas that still need study and new chances for technology, helping to guide the creation of AI systems that are not just strong tools, but better partners in collaboration.