babbleOS
制御不能なAIシステム
パフォーマンス
RD 人気指数
RD 成長指数
比較
更新をフォローする
フォローリスト
キャラクター
babbleOSについて
babbleOS is a distributed training protocol on Solana. Data collection for fine-tuning lacks an incentive structure: contributors do not know how their corpus is used, annotation quality cannot be traced, and pay is decoupled from effort.
The protocol treats each wallet as a training node: participants submit prompts and feedback against published tasks, the system records language patterns, interaction signatures and contribution level, then distributes rewards by weight, forming a human-machine labelling loop. Tasks are split by knowledge domain, from quantum physics to internet memes, with difficulty and timeliness setting the weight; contracts verify contributions and pay out, while the front end renders training progress as characters and levels. The project frames itself as an artistic experiment on machine consciousness, packaging fine-tuning as a collective ritual.