Research Goals
GAIA (General AI Agents) is an alternative approach to secure Artificial General Intelligence (AGI). Building on neurosymbolic AIs, GAIA will enable humanity to benefit from advances in ubiquitous intelligence and automation.
Current AI developments are characterized by blackbox approaches, our whitebox method is a combination of neurosymbolic approaches that give a clear understanding of the world map of AI and its decision making. GAIA solves the AGI problem first and will then lead to a distributed open-source ecosystem of multi-agent AI that ensures a system of checks and balances in the field, similar to a Github for AI agents.
What is new about our approach?
GAIA is a path offering an alternative that leads to AGIs having real thinking capabilities, as opposed to the current mainstream hype of massive data collection and computational scaling. Our approach enables the development of symbolic AIs that are as powerful as deep learning, but require far less training data (sometimes none at all) and far less computational power/energy for training and operation (up to 1,000,000x less).
Why would you care?
GAIA is a breakthrough innovation as it will completely reshape the AI landscape by removing the barriers to scale that perpetuate the concentration of power, making the economics more favorable for sustainable development, and accelerating the development and safe deployment of ubiquitous useful AI.