Language Games as the Pathway to Artificial Superhuman Intelligence

2025-01-31

AI Summary

Large language models (LLMs) tend to fall into mode collapse during data cyclic regeneration, hindering emergence of superhuman intelligence. Method: We propose the “language game” paradigm—a human-AI collaborative, open-ended evolutionary ecosystem grounded in three core mechanisms: role fluidity, reward diversity, and rule plasticity. The framework integrates multi-agent systems, reinforcement learning–based feedback, dynamic constraint modeling, and iterative retraining. Contribution: This work introduces the first language game framework explicitly designed to break LLMs’ data regeneration closed loop. Empirical results demonstrate sustained generation of novel, high-quality data; significant improvements in cross-task generalization and creative reasoning; and superhuman growth trends across multiple open-ended exploration benchmarks—including AIME, MMLU-Pro, and GAIA. By enabling continuous, adaptive co-evolution of language, cognition, and task structure, our approach provides a scalable pathway toward artificial superintelligence.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Humans and AI: Game Design — Procedural Content Generation & Storytelling

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

The evolution of large language models (LLMs) toward artificial superhuman intelligence (ASI) hinges on data reproduction, a cyclical process in which models generate, curate and retrain on novel data to refine capabilities. Current methods, however, risk getting stuck in a data reproduction trap: optimizing outputs within fixed human-generated distributions in a closed loop leads to stagnation, as models merely recombine existing knowledge rather than explore new frontiers. In this paper, we propose language games as a pathway to expanded data reproduction, breaking this cycle through three mechanisms: (1) extit{role fluidity}, which enhances data diversity and coverage by enabling multi-agent systems to dynamically shift roles across tasks; (2) extit{reward variety}, embedding multiple feedback criteria that can drive complex intelligent behaviors; and (3) extit{rule plasticity}, iteratively evolving interaction constraints to foster learnability, thereby injecting continual novelty. By scaling language games into global sociotechnical ecosystems, human-AI co-evolution generates unbounded data streams that drive open-ended exploration. This framework redefines data reproduction not as a closed loop but as an engine for superhuman intelligence.

Problem

• Large Language Models
• Pattern Breaking
• Superintelligence Development

Innovation

• Multi-role Interaction
• Diverse Reward Systems
• Adaptive Rule Dynamics