OpenAI Leak Suggests AI Could Run Companies by 2030

OpenAI Leak Suggests AI Could Run Companies by 2030

The leak of the Strawberry project, previously known as Q* star, has reignited alarms regarding the actual timeline for Artificial General Intelligence (AGI). The message, presented on August 13 at 19:00, argues that by 2030 an AI will be capable of running a company independently. The implication drawn from this statement is stark: if the average worker becomes unnecessary, the system no longer needs people it currently deems obsolete. This date is not a minor detail. It marks the horizon where, according to this perspective, automation would cease to be a tool and become a substitute.

What Is AGI and Why It Is Confused with ChatGPT

The underlying confusion is technical. An AGI is not just a search engine with a better interface. The difference has been explained through a simple analogy: ChatGPT is like a Google that, instead of returning links, reads them and provides a human-like summary. The interface is deceptive. Behind it lies no understanding, but a highly complex statistical algorithm predicting the next word. That is not reasoning, even if it appears so.

AGI requires a different leap. It would need to self-program, master disciplines it was not trained on, possess something akin to self-awareness, and learn new subjects without supervision. Today, it remains a concept. No machine meets these criteria. Therefore, the debate is not whether ChatGPT writes well; it is whether the next step exists or is merely hype.

The Physical Limit: Energy and Water

The strongest argument against the triumphalist narrative is not philosophical, but material. AI requires vast amounts of energy and water for training and operation. Data centers consume resources that do not scale at the pace demanded by the hype. A seven-year horizon for an AI to manage an entire company clashes with this constraint. It is not a software problem; it is a physics problem.

Some argue that real pogre lies not in language models, which may already be near their ceiling, but in technologies that do not yet exist. AGI would require a different architecture, not more parameters. According to this view, the most likely scenario is that everything remains a very powerful tool. Useful, yes. Autonomous, no.

Universal Basic Income and the Illusion of Perpetual Leisure

The implicit promise is that if machines work, someone will pay the bill. Universal Basic Income (UBI) appears as the sweetener making the process digestible. The objection is obvious: if no one has a job, no one buys the phones robots manufacture. Consumption relies on wages. Without wages, the cycle breaks.

Distrust toward this promise is widespread. A tycoon does not support the population so they can watch Netflix. The history of automation is not one of liberation, but of power concentration. Mechanization of agriculture was not offered; it was imposed. The result was not more leisure, but exodus and dependency.

Jobs That Survive (For Now)

The practical response from part of the analysis is education. Programming languages and robot maintenance appear as the last refuges. Remote work, consolidated after the pandemic, is perceived as the only transferable asset. Continuous learning is not a choice; it is a race against the calendar.

The problem is that AIs also program. They are already improving themselves. This refuge has an expiration date. Human advantage, if it exists, lies in rapid adaptation, not static knowledge. Learning how to learn sounds like a textbook slogan, but it is the only strategy that does not depend on the model stagnating.

---

The exact point where the analysis stalls is the same as at the beginning: no one knows if the leap to AGI is a matter of years or decades. The Strawberry leak does not resolve the doubt; it postpones it. Meanwhile, the 2030 date remains there, a reminder that someone is already setting deadlines.