Giving Plants a Voice: Designing a Conversational Agent

As artificial intelligence becomes more ingrained into everyday elements of life, our team began to question how AI could be applied to the natural world and helping us better understand plants. As such, we formed this study to in a sense “Give voice to these plants,” and help us learn more about them. Our study combines AI through conversational agents with plant imaging software of common apps and identifiors. Our goal was to merge both into a physical prototype that still allows users to engage with the natural world (without the distraction of an iphone), while also being able to converse with the gadget naturally and recieve in depth, personalized answers that were unique to their specific plant’s needs and conditions.

My process began with in-depth research into plant identifiers and conversational agents, where I found that the main issues in both systems centered on user experience and functionality. Conversational agent users questioned the invasiveness of the technology and wanted clearer sensory cues to indicate when devices were active, while plant identifier users doubted the accuracy and limitations of classifications.

In response, my team and I explored how these systems could merge and how conversational agents might give voice to plants. I developed these ideas through sketches, renderings, low-fidelity paper mache and cardboard prototypes, and eventually a fully 3D-printed final prototype.

Together, our four-person team tested the prototype through Wizard of Oz and autobiographical testing, allowing us to identify improvements, refine interactions, and explore alternative user experiences.

Methodology:

My initial ideation began with 30 thumbnail sketches conceptualizing various 3D models and conversational agents. From these, the designs were narrowed down into 10 concepts to develop further.