Skip to content
Sam Altman · Technological singularity · OpenAIEpisode 145 · 5 August 2026 · 31:10

Has the Singularity Already Begun? Who Will Control Superintelligence?

Central question

What can actually be considered the beginning of technological singularity, and who will control access if AI truly approaches superintelligence?

What you take away

Separate the singularity debate into model capabilities, economic effects, AGI, ASI, and control instead of searching for one date or one magical indicator.

Main threads

What to watch for

1Separate growth in model performance, autonomous action, and real economic change; these are different signs of acceleration.
2Do not infer singularity from one release: watch several independent lines—agents, science, robots, and distribution of access.
3Note where a powerful capability has become an everyday tool and where it remains a demonstration without mass adoption.
4In every scenario, ask who owns the system, who sets access rules, and whether a person or state can reject its decisions.
Signals to track afterwards
The transition of AI agents from answering to sustained independent work.
Scientific results that a model helps produce rather than merely summarize.
The connection between digital intelligence, robots, and the physical world.
The concentration of compute and access rights among a small number of companies and states.
Most useful for
people trying to understand AGI and singularity without hypefounders and investorsAI policy and safety specialistsleaders planning long-term strategya broad audience following AI development

Key takeaways

00:00A Singularity Claim Is a Frame, Not a Measurement

Sam Altman’s statement that humanity is already inside the singularity describes a sense of acceleration. It does not provide one testable criterion for when the era began.

01:20A Super-Powerful System Immediately Raises the Question of Power

The image of a “machine god” and the idea of AI freedom for everyone lead to the same fork: whether the system will be centralized or access distributed among people and organizations.

05:20Singularity Means a Loss of Predictability

The term refers not simply to a powerful model, but to a point after which technological development accelerates and becomes difficult for humans to understand.

12:20ANI, AGI, ASI, and Singularity Are Not the Same Thing

Narrow intelligence, general intelligence, superintelligence, and irreversible acceleration describe different levels. Mixing the terms creates an illusion of a more precise forecast than actually exists.

16:15There Is No Single Measurement

A model test, compute volume, or user count does not prove singularity on its own. What matters is a combination of changes in capability, autonomy, and the real world.

19:05Acceleration Can Coexist with Ordinary Life

Tools can change quickly while infrastructure, organizations, and everyday processes change slowly. A person can live inside technological acceleration and barely feel it in daily life.

22:22A Miracle Quickly Becomes an Interface

A capability that recently seemed impossible becomes an ordinary button within months. The sense of a breakthrough therefore keeps disappearing as people adapt.

23:45The Next Stage Is Tested Through Agents, Science, and Robots

The forecast connects further acceleration with systems that independently move work forward, help produce discoveries, and enter the physical world.

What this episode is about

Sam Altman says humanity is already inside the singularity. But the term combines several different ideas: accelerating models, the arrival of AGI, possible superintelligence, and the point at which technological development becomes impossible for people to understand.

The Singularity Cannot Be Reduced to One Powerful Model
The discussion separates four levels: narrow AI, artificial general intelligence, artificial superintelligence, and technological singularity. These concepts are often blended together even though they describe different thresholds. A model can outperform people in selected tasks and still not be AGI. Superintelligence implies a different scale of capability, while the singularity describes a broader break after which the pace of change becomes difficult to predict.

There Is No Single Meter for the Singularity
There is no dashboard showing that humanity has reached 63 percent of the singularity. Some people look at model quality; others focus on autonomy, scientific discoveries, robots, or economic impact. Everyday life may still look familiar for a long time: people visit the same stores, use the same roads, and continue many old processes. A major technological shift does not have to transform every ordinary activity at the same moment.

Yesterday’s Miracle Quickly Becomes an Ordinary Tool
Claude Opus 5 and other new models show how quickly the sense of normality changes. A capability that seemed almost impossible a few months earlier becomes another feature and stops feeling like evidence of a historic break. A person can therefore live inside acceleration without feeling it. Altman connects the next stage with agents, scientific discoveries, and robots—systems that do more than answer questions and begin moving work forward on their own.

The Central Question Is Not the Label, but Control
The debate about the singularity quickly becomes a question of ownership and power. Will an extremely capable system be concentrated inside one company or state, or will access be distributed among people, countries, and organizations? Broad access creates freedom as well as new risks. Centralization promises control but grants enormous power to the system’s owner. The more important task is therefore not proving that the singularity began on one particular day, but understanding who will make decisions about AI’s development and who will receive access to its capabilities.

The debate over whether singularity has begun matters less than observing concrete signs of acceleration and the distribution of power. The central question is not the name of the era, but who will be able to use the system and who will set its rules.

Episode transcript

The episode is in Russian; below is an English reading guide to the transcript (the full EN transcript is a machine translation). Voice matching applied to 50 segments: 50 identified, 0 mixed, 0 marked with ✓, and 0 unresolved.

Loading…