ASIASIApplied AI systems · Research · Architecture

HS-AGI · Working hypothesis

Not an AGI achievement claim

Human-led.
Machine-extended.

Could a deliberately structured human–AI system become more capable across a broad range of tasks than either participant working alone?

Beyond the prompt

Towards shared capability.

HS-AGI means Human-Synergistic Artificial General Intelligence. It is a working research hypothesis, not a released product, a new model or a claim of achieved AGI.

We are interested in complementary strengths, shared context, memory support, tools, feedback and explicit human authority. Our ambition is to make useful collaboration accessible beyond specialist labs.

Human alone

Establish a strong human baseline for each task family.

AI alone

Establish a strong AI baseline with clearly declared tools and resources.

Human + AI

Test whether a deliberately designed collaboration adds value beyond either baseline.

Use repeated tasks, independent scoring where possible, and explicit accounting for time, tools, cost and assistance. Beating both baselines on one task family is evidence about that family—not proof of general intelligence. A failed prediction constrains the tested design, not every possible form of human–AI collaboration.

No claim of consciousness, universal superiority, guaranteed social benefit, scientific priority or imminent public availability.

Discuss the research question →