Emergent Hyperdimensional Scientific Discovery System

Inspiration

Current scientific process and ideation is hard, costly and time consuming, so I decided to change that using this software

What it does

It encodes scientific hypotheses, experiments, and data as high-dimensional contextual embeddings.

Hypotheses interact dynamically, producing emergent clusters representing potential laws.

Experimental or simulated feedback loops reinforce or weaken these clusters causally.

Laws emerge as attractor states, evolving continuously beyond fixed axiomatic limits. T The system self-mutations embedding structures and feedback parameters to innovate autonomously. Cross-domain integration fosters novel interdisciplinary discoveries.

Finally, human-readable laws are extracted from the emergent embedding landscape.

EHSDS offers a revolutionary, non-symbolic, adaptive, and experiment-grounded scientific discovery engine poised to accelerate breakthroughs beyond traditional formal logic systems.

How we built it

I built it completely through the use of AI Studio and gemini-3-pro-preview.

Challenges we ran into

There were some bugs we ran into. At first, it wasn't generating any novel ideas, but we managed to solve that through Gemini-3-pro-preview's reasoning capability. Second problem was we kept running into out of token errors, but there was nothing we could do about it.

Accomplishments that we're proud of

I've tested it using a seed dataset containing 200 ideas, it managed to come up with more than 30 new and novel ideas that held up rigorous testing.

What we learned

I've learned that AI is much more capable than just answering normal questions, only if you use it with the right tools and workflow.

What's next for Emergent Hyperdimensional Scientific Discovery System

If I manage to win this hackathon, I will use the money to use it to come up with more ideas

Built With

• gemini
• geminiapi