Sam Altman lowers the bar for AGI

For years, OpenAI sold artificial general intelligence as a quasi-mythical technological singularity capable of automating human labor and uplifting civilization. Now, Sam Altman is telling everyone to take a deep breath.

Speaking at The New York Times DealBook Summit, the OpenAI CEO significantly dialed down expectations for what happens when his company actually hits that milestone. Altman remarked that the world will hit AGI sooner than most people think, and it will matter much less.

Moving the goalposts

Altman explained that the world will mostly just carry on as normal once AGI arrives. In his view, many of the existential safety concerns voiced across the industry will not materialize at the moment of AGI, because the leap from AGI to what he now calls superintelligence is going to take a long continuation.

According to Altman, we could see early iterations of AGI as soon as 2025 running on existing hardware. These systems will look like an assistant capable of taking on complicated human-level tasks and coordinating different tools to finish them. The Verge also reported that OpenAI might simply weave together its existing large language models and slap the AGI label on the final package.

Meanwhile, the jaw-dropping societal disruption that OpenAI’s charter once attributed to AGI has been quietly rebranded as superintelligence, a phase Altman recently estimated could still take a few thousand days.

Altman also shared fresh numbers, noting that ChatGPT has now crossed 300 million weekly active users. He compared the proliferation of AI to the invention of the transistor, arguing that reasoning models will soon be cheap commodities built into everything.

The Microsoft escape hatch

Lowering the bar for AGI is not just about soothing public panic. It is a calculated corporate maneuver.

OpenAI is currently in the middle of a delicate pivot toward becoming a conventional for-profit enterprise. Buried inside its multi-billion-dollar partnership with Microsoft is a critical contractual clause: the revenue-sharing and exclusive IP licensing terms effectively expire once OpenAI officially declares the arrival of AGI.

If AGI is defined as a god-like intelligence that automates all human labor, OpenAI might never trigger that clause. But if AGI simply means an ensemble of models reliably chaining tools together to finish a workflow, OpenAI can declare victory comparatively soon, pull the escape hatch, and free itself from Microsoft’s commercial grip.

Why it matters

For everyday users and developers, this reality check is actually good news. You do not need to brace for overnight societal collapse or immediate mass technological unemployment in 2025. What Altman is describing looks less like a sentient sci-fi mind and more like a very competent, modular agent capable of saving you a few hours of screen time.

For the AI industry, however, the linguistic gymnastics are hard to ignore. We think Altman is pulling off a remarkably clever two-step here. By uncoupling AGI from superintelligence, OpenAI can placate regulators worried about frontier doom while simultaneously positioning itself to cash out on its corporate independence from Microsoft.

The downside is trust. When the most prominent company in the space moves the goalposts to fit contract deadlines and PR cycles, AGI risks becoming just another diluted marketing buzzword. If achieving AGI feels like an anticlimactic software update, the industry may find that hype is a resource you can only burn once.