🎮Gamifying AGI research
Last updated
Last updated
HELP is research lab focused on synthetic dataset generation. Virtual environments can provide rich, diverse synthetic datasets encompassing various modalities. The skills acquired through gaming, including strategic thinking, real-time decision-making, and adaptability, are highly transferable to other domains and represent an invaluable synthetic dataset for training models. HELP’s ultimate goal is to create generalist AI agents capable of understanding and performing tasks across various 3D virtual environments, without needing access to game source code or specific APIs. This system would be considered a good start to the emergence of general intelligence.
Like the ARC-AGI 1M$ Prize, a competition designed to measure progress towards artificial general intelligence. HELP is designed to create and measure an agent's ability to efficiently acquire new skills and solve novel tasks outside its training data in a gaming setup.
