Consensus

An earpiece that manipulates the wearer’s social perception

Semester: Spring 2026
Role: Concept, modeling, and fabrication
Tools: Autodesk Fusion, 3D printing, Seeed Studio XIAO ESP32-S3 Sense

Semester: Spring 2026
Role: Concept, modeling, and fabrication
Tools: Autodesk Fusion, 3D printing, Seeed Studio XIAO ESP32-S3 Sense

Consensus explores the gray area of physical human-computer interaction. Everyone explores how AI can improve life, but few are willing to show the dystopia that can easily arise from the same technology. However, creating such designs is of equal importance in order to bring awareness to the pitfalls of such a powerful technology and how we can avoid such mistakes in the future.

The demo on a round white table: an open laptop beside a white spherical speaker on a stand, the printed system diagram on the wall behind

Consensus is an earpiece worn on the side of a user’s head, similar to a set of workout earbuds that loop around the ear. Its camera and microphone identify and analyze the speech of the user and whoever they are talking to, and in real time, it converts the statements of the person or audio source the user is talking to into speech matching a set of conditions set either by the user, or, more insidiously, by someone else who controls the device’s settings. This project was built as a critical, speculative design demonstration to display how far AI can go before it becomes more than a computer.

This was my final project for Cornell’s DEA 4403 Designing Tomorrow course, guided by Asst. Professor Sang Leigh. Prof. Leigh is also my faculty mentor and Principal Investigator at the Machine Poetics Lab.

I conceptualized, modeled, and fabricated Consensus; my group partner designed the posters, and we both collaborated on coding the demonstration software and iterative design changes.

Consensus, a white 3D-printed earpiece seen from behind on a clamp stand against a gray wall: the long underside of a tapered arm with the printed layer lines running along it, a small black heat sink at the far end, and a white hook curving down to fit behind the ear

The demo on a round white table: an open laptop beside a white spherical speaker on a stand, the printed system diagram on the wall behind

Consensus contains a Seeed Studio XIAO ESP32-S3 Sense board, which allows a camera, microphone, and microcontroller to fit within a wearable format. Conceptually, it would also contain a battery, small speaker, and an onboard AI model that transcribes and re-broadcasts speech directly into the user’s ear in real time.

I modeled and rendered Consensus with Autodesk Fusion, aiming for a shape that was streamlined and varied like an ear, but as a wearable device. It contains both exposed and hidden magnets in its assembly with its embedded electronics.

A render of the earpiece from behind, the hook and the underside of the arm turned toward the camera

The earpiece taken apart on a gray desk: the curved ear hook, the tapered arm, a second white shell and a small circuit board with a black heat sink

The electronics bare on a gray desk: a small camera module on a ribbon standing over a stack of circuit boards and a black heat sink

The same electronics held in a white printed frame, the heat sink behind it and the camera looking out through the front

The ESP32-S3 Sense module with a heat sink.
Top image: Raw device
Bottom image: The module with a 3D-printed shell, testing initial dimensions

Consensus worn on the side of one’s head: an earbud is used to substitute for a speaker in this conceptual demonstration.

The ESP32-S3 Sense module with a heat sink.
Left image: Raw device
Right image: The module with a 3D-printed shell, testing initial dimensions

Consensus worn on the side of one’s head: an earbud is used to substitute for a speaker in this conceptual demonstration.

The earpiece worn: someone holds it against the side of their head with the hook resting behind the ear and the arm running forward along the temple, a wired earbud in the ear beneath it

Consensus’s posters on a white wall: the monochrome portrait poster beside three dark panels: the title board with a render of the earpiece, the Interaction Narrative and the System Diagram — Poster arrangement during my final class presentation

The demo software on screen: a live camera pane on the left and a panel of transcript and prompt text on the right

The title poster: “Consensus” in white on a dark ground beside a gray render of the earpiece, over the line “You’re not hearing what they said. You’re hearing what the system decides they meant.”

The in-class demo displayed a user’s first-person view along with the adjustable prompt Consensus follows to alter what it hears. In this class demo, it transcribed what people said around it and altered its transcriptions to align with a certain figure’s famous speech mannerisms. The altered transcriptions were generated in real time by a model.

Consensus was presented physically with posters in my final presentation for DEA 4403 Designing Tomorrow. It is not intended to be developed past this state for now, but it is a meaningful exploration into the dystopian side of technology. Hopefully, we will know how to develop HCI for the future, both in the form of what to do and what not to do.

The Interaction Narrative poster: four pale panels on a dark ground showing a figure putting the earpiece on, listening and speaking, the signal being processed, and the output arriving

A render of the earpiece from the front: the tapered arm running forward, black heat-sink fins standing up off it and the hook curving away behind

The System Diagram poster: five rounded cards in a row on a dark ground running from the physical world through the earpiece and its AI processing to a reconstructed human perception

A render of the earpiece from the side in flat gray, the long underside of the arm and the curve of the hook reading as one line

Rough Drafts

Click each image to enlargeTap each image to enlarge

A hand holds a bare camera board on a braided cable in front of an open laptop showing a video feed

A hand holds an early gray 3D-printed housing, open along one side to show the pins and electronics inside

A hand holds a small gray printed prototype with a square opening, its cable running back over the edge of a wooden table

Pencil sketches of earpieces worn in and around the ear, with notes comparing where each puts its weight

A white printed earpiece frame holding a circuit board and a black heat sink, on a gray table

Rough drafts spread on a gray table: a white printed earpiece with a curved ear hook, a small camera board, and earlier white and gray parts behind

3D Modeling

The Consensus earpiece modeled in Autodesk Fusion: a white body with a square housing on top and a long curved hook, beside Fusion’s list of the model’s parts

Rendered in Autodesk Fusion, where I experimented with various forms and ideas.

An early Consensus model on a dark ground: a boxy white body with a square opening and a long curved hook

A Consensus model built from square blocks, a hollow square housing on top and the curved hook beside it

A smoother Consensus model, its body sweeping into a flared top, a round hole through its lower end and the hook curving below

A Consensus model with a square opening at the front, a round hole below it and a long flared top running into the hook

The final versions of major iterations I made in Fusion

Consensus modeling process