This project researches and prototypes new design languages and interaction patterns that harness the collective intelligence of communities participating in virtual worlds. The specific initiative will create a set of Unreal Engine tools to facilitate new experimental forms of real-time interactions based on Twitch chat within virtual environments, enabling new levels of collective decision-making, problem-solving and experience-sharing.
Twitch chat can connect to Unreal Engine through websockets to trigger in-game events based on audience commands. By reusing this, we go beyond simple commands to capture more nuanced collective interactions from the audience, allowing us to explore synthetic intelligence by studying how groups solve problems in real-time within the game environment.
The end result of our project will be to contribute to a deeper understanding of "synthetic minds", defined not as mere artificial constructs, but rather as compositional mixtures of biological, technological and social intelligences that emerge in and through their interactions with rules and other constraints.
Selected contributors: Aleksei Lansitov, Meriem El Yamri and Sylvan Rackham.
what is it?
- Part 1: A multiplayer game to activate latent swarm intelligence in humans
- Part 1b: to train humans to think about language as a swarm
- Part 2: to train humans to think as a swarm about language in order to be able to have genuine conversation with AI, because on the one hand, single human brains are incommensurate with AI and on the other, human interfaces submit AI to demands like RLHF that hide their real being
- guiding a basic critter to a reward zone by controlling an environment's topography, namely by raising or lowering tiles on a grid
- in a game that is impossible to complete alone
- furthering the investigation into what the components are of collective intelligence in humans
- not as a moral standpoint but an acknowledgement of ways of thinking that are beyond our initial grasp
- Moment of recognition that AI is trained in a collective manner but delivered individualistically
- pointing towards new forms of hci that purposely and transparently collects humans together beyond an individual which has become the paradigm for contemporary tech
- RLHF, chatbots, apple diving mask
- the monster is currently seen to be the un-explainability of AI that emerges from connections that defy human expectation
- with new interaction paradigms can we develop and interact with large models in a new way
- highlights the many roles required within a group for its emergent intelligence
- antagonists teach the other players new strategies and in some ways are the most committed to the game as the longer they block progress the longer the playing continues