Specialists may lead.
The system may follow.

A few names for a big idea.

Specialized Superintelligence / SSI

A system that reliably surpasses the best human experts and institutions, using their best tools, across a substantial part of an important field.

General Super Intelligence / GSI

A system—or a coordinated group of systems—that combines broad reasoning with specialist strength and reliably surpasses the strongest human teams across virtually all cognitive work.

Systems helping build better systems

SI helping improve the software, chips, algorithms, data, experiments, and evaluations behind the next generation. The technical term is recursive improvement.

The singularity

A possible turning point when discoveries, better systems, and better infrastructure help progress accelerate further.

These are our working definitions. The pathway is a possibility to explore, not a claim that sustained SSI or GSI has been achieved here.

Specialists could work as a team.

Think of a difficult research project. One specialist might be excellent at mathematics, another at software, and another at biology. A more general system could help divide the work, compare their answers, and keep the project moving toward the right goal.

A possible pathway, not a prediction or a record of achieved progress.

Good teamwork is not automatic, even with powerful specialists. Their answers might clash. They might make the same mistake. Somebody—or some system—still has to notice the problem and check that the final result solves the original task.

Brilliant at one thing is not brilliant at everything.

A system could outperform experts on one demanding problem and struggle with an unfamiliar job next door. General intelligence adds the ability to work across fields and adapt. GSI raises the bar again: beyond the strongest coordinated human teams.

These steps describe what a claim would need to establish, not ratings we are awarding here. Whether the field is software, mathematics, biology, robotics, or something else, an extraordinary result needs to hold up repeatedly and under independent scrutiny.

Progress still needs power, people, and patience.

Better systems could help design more efficient software, better chips, and more useful experiments. Those improvements could make the next generation more capable. The whole research and engineering effort might advance—not just one model rewriting itself.

But a proposed chip still has to be manufactured. An experiment still has to run. A facility still needs power. And a clever-looking answer still needs to be right. Moving faster is exciting; making dependable progress is the point.

The future should be good for people.

What would make this future worth building? More than a higher score. Better tools should help people learn, create, solve problems, and choose how they live. Human judgment and relationships do not become less important because a machine becomes more capable.

None of that follows automatically from more capable technology. It takes good decisions, institutions people can trust, and a clear view of who benefits. That is the future we want to work toward.