Consensusmachine
AI inzetten om consensus te bereiken
ProjectDe Consensusmachine is een interactieve AI-installatie die mensen helpt bij het bereiken van consensus over een controversieel onderwerp. Met dit experiment onderzoekt de Werkplaats Responsible IT het idee van een onafhankelijke AI mediator, die de oer-Hollandse praktijk van het Polderen ondersteunt.
DThe Consensus Machine - also known as the Polder Model - consists of a large interactive touch screen that bystanders operate during a group discussion. The AI starts by writing a stimulating text that appears on the screen. Participants in the discussion use the touch screen to select the phrases they want to tone down or reinforce. The machine immediately displays alternative wording on the screen, and bystanders choose an option acceptable to all. The AI rapidly rewrites the text of the proposition in the spirit of the chosen alternative, and the process begins again. The participants let the AI rewrite the proposition until everyone can live with the result.
Co-creation: staff, students and researchers
The Consensus Machine was developed by HvA students and researchers from the Studio Responsible IT ("de Werkplaats"), together with the HvA's Department of Digital Strategy and Information Policy (DSI) for the opening the 2023-2024 academic year.
Read all about the project in the PDF at the bottom of this page.
The Consensus Machine is a technological experiment through which the Responsible IT Workshop makes tangible the idea of the independent AI mediator. In this experiment, the creators used the SURF Values Guide for Digitization as a framework in which GPT4 writes the alternative formulations. The experiment raises the question to what extent technology is neutral and value-free. Does a human really have control over a decision-making process, or is she unconsciously steered in a certain direction by the creators of the AI?
Can a technology such as AI that creates polarization also be used to bring people together? Potentially, the Consensus Machine is a tool that can be used anywhere to achieve and experience consensus. The experiment shows how making research connects education to creativity, societal issues and Human-AI interaction.
The Studio Responsible IT Amsterdam Workshop (RITA) researches software that is good for people, society and the planet. The young research team of programmers, data scientists, designers and ICT students experiments by creating working algorithms, datasets, models and software. These products are developed with end users whenever possible.
All software produced is open-source, and available for anyone to use. For questions about the workshop, or the research projects, please contact Yuri Westplat, head of the Studio Responsible IT (y.westplat@hva.nl).
