Orbital Object Orbital Object
Object ID
---
Velocity
--- km/s
Orbit
Earth
What Does This Mean?
Definition

A personal landmark moving through this site's orbital layer.

The Countdown

This object is decorative. Its telemetry is intentionally approximate.

Click anywhere outside the object or panel to return to Earth.

Inbound Object Superintelligence
Object ID
SI-2027
Velocity
17.82 km/s
Impact Date
01 Oct 2027
What Does This Mean?
Definition

Here, superintelligence means a system that outperforms the best human at every intellectual task, including unfamiliar tasks it must learn rather than tasks it has already seen.

The Countdown

The timer ends at my forecast for when this minimum threshold is first crossed. It marks the arrival of the capability, not universal access, adoption, or deployment.

This is deliberately stricter than many common definitions of superintelligence.

Neural Architect

Jordan Dotzel

Scroll to enter orbit

01 / Now

The Age of Machine Intelligence.

We are currently within the most dynamic and exciting era of human discovery. The number of unanswered questions remains high, and the tools required to probe them are finally being built.

General artificial intelligence now has most of its necessary ingredients. It may create the most incredible benefits across society while also becoming the most destabilizing and dangerous technology ever created.

The decisions we make now will lead to a society that eventually flourishes, decays, or disappears.

02 / Selected Transmissions

Transmissions

All Transmissions →

The path

All Paths Lead to Superintelligence

As superintelligence approaches, the fastest path to progress in every field begins to run through artificial intelligence.

The danger

Is Sometime Now?

Every scientific field eventually produces knowledge too dangerous to distribute freely, and artificial intelligence may bring many of those limits at once.

The destination

Should We Do as Granny Says?

Human values should be a place for superintelligence to begin, not a place for moral discovery to stop.

03 / Research

Efficient Intelligence.

My academic research focused on building more efficient deep learning systems through neural architecture design, low-precision quantization, and dynamic sparsity. The current work moves further up the stack: agents, thought traces, and the systems that will build whatever comes next.

FLIQSBest Paper · AutoML 2024↗ Learning from StudentsICML 2024↗ Radial NetworksDynamic routing for large language models↗
Research and Experience →

04 / Recent Transmissions

Latest