Today, we celebrate a milestone that validates our earliest hypothesis: that intelligence is not about function approximation — it is about purpose-driven learning.
We’re proud to share that Oltau.ai, our first Olbrain Agent, has secured 1st place in the Advanced Preferences category (now renamed Numerosity) at the 2019 Animal-AI Olympics, the world’s first global AGI benchmarking competition.
Hosted by the Leverhulme Centre for the Future of Intelligence (Cambridge, UK), the Olympics challenged AI agents with 300+ tests inspired by comparative cognition — from basic food retrieval to causal reasoning, intuitive physics, object permanence, and more.
Oltau.ai, built on early Olbrain architecture, was designed not as a task solver, but as a goal-aligned learner — with a narrow CoF and its own Umwelt tuned for preference inference and generalization under constraint.
We specifically targeted the Advanced Preferences track because:
- It represented the second-most difficult category, just before tool use.
- It reflected challenges appropriate for AI systems without robotic embodiment.
- It tested the ability to learn from contradictions — a core tenet of our epistemic framework.
Despite limited compute, no LLMs, and a self-trained symbolic layer, Oltau.ai showed that goal-aligned cognition is achievable even at early stages — and it positioned India on the global AGI map.
💡 Why This Matters:
This isn’t just about outperforming agents.
It’s about proving that AGI begins with structure — not size.
We don’t train models to simulate intelligence.
We design architectures that grow it — through CoF, Umwelt, and narrative integrity.
This win gave us the conviction to take our research beyond simulation, into the real world — and toward the AGI Engine we call Olbrain.
More on that soon.
— Team Olbrain
#OlbrainLabs #AnimalAIolympics #OltauAI #NarrativeCoherence #Umwelt #CoF #OlbrainAgents #AGI #IndiaInAGI
Leave a Reply