The current study looks to assess how effective ChatGPT can be in helping students learn and complete statistical tasks through understanding how students utilize it, what its capabilities are, and the scaffolding processes it can perform. The software is readily available and low cost which could make it a valuable tool for both students as well as faculty during statistics education. As an ongoing project a total of 15 participants will be recruited for the study but only three are discussed. Over the course of three weeks participants were asked to complete a statistical task over Zoom each week utilizing the software in any way they thought helpful in individual breakout rooms. After each task participants provided the transcript of their session and were interviewed about what occurred during the task which were the sources for the thematic coding. After submission of the transcripts from the final task themes were compared across participants to draw an overall narrative of their experiences. First, it was found that the software responds best when entered prompts contains prior output information and the software is given the task in its entirety as a frame of reference. When examining for the various forms of scaffolding the software related most well to conceptual and strategic while struggling with discourse as well as procedural. Lastly, the software showed limitations in its ability to locate and utilize desired statistical knowledge which heavily impacted one of the tasks. From these findings best practices for use of the software are recommended and ideas for implementation are discussed.