This panel will discuss grand challenges in visual analytic (VA) systems for making the research field more rigorous, valuable, and impactful. Our panelists represent academic and industrial researchers with extensive experience in VA with varying expertise from theories to applications. The panelists will discuss their perspectives on challenging questions such as: How can we strengthen the theoretical underpinning of VA? Do we value too much on novel yet often complex visualizations? How can we enable tight integration of AI/ML into VA systems? What are opportunities for more rigorous evaluation strategies? How can we derive transferable and general knowledge from ad-hoc VA systems? How do we motivate deployments and open-sourced demos? We anticipate this panel will function as a way of airing and calling for action on those challenges.