OpenAI CEO Sam Altman is signaling that the pace of AI development inside leading labs may be accelerating faster than expected. In an interview at the Express Adda event, he said OpenAI is already using its own AI technology internally to speed up research, and he connected that acceleration to a sharper outlook on AGI and superintelligence.
The message is not simply that stronger models are coming. It is that the tools used to build those models may now be improving the building process itself.
OpenAI is using AI to accelerate research
Altman said OpenAI is already applying AI internally in ways that make development move faster. According to the source article, he described AGI as "pretty close" (41:07), while superintelligence is "not that far off" (41:53).
That framing matters because it shifts the discussion from a distant theoretical milestone to a nearer operational concern. If AI systems are helping researchers move more quickly, then progress is not only about one model being more capable than the last. It is also about the research cycle becoming faster.
Altman described that pace directly: "It's going to be a faster takeoff than I originally thought and that is stressful and anxiety-inducing," he said (38:45).
The source article also says Altman suggested that OpenAI already has models that go beyond what is publicly available. From the lab perspective, he said, "the world is not prepared" for what is coming.
AGI and superintelligence are becoming near-term concerns
Artificial general intelligence, or AGI, is presented in the source article as a central part of Altman's comments. He did not describe it as a remote idea. He said it is "pretty close".
Superintelligence was placed close behind. Altman said it is "not that far off", which makes the concern broader than the arrival of one new product or one public model release.
The practical implication is that public expectations may lag behind what AI labs believe is approaching. If the most capable systems inside a lab are ahead of what users can access, then the public view of AI capability may be incomplete.
OpenAI recently said that its new coding model Codex 5.3 was co-developed by the model itself. That detail fits the broader theme of AI being used as part of the AI development process.
The software job is changing first
Altman also connected AI progress to the future of work, with software development as one of the clearest examples. His point was not that software developers disappear entirely. Instead, he said the way the work is done is changing sharply.
"The way I learned to write software is now effectively completely irrelevant."
He added that there will still be software developers, but that "writing, you know, C++ code by hand, that's over" (54:34).
That is a direct challenge to older assumptions about programming skill. If AI systems can take over more of the manual production of code, then the value of a developer may move away from hand-writing every line and toward deciding what should be built, how it should behave, and whether the result is correct.
The source article does not present this as a uniform effect across all work. Altman said, "Big categories of jobs AI is just going to completely obsolete," while others will barely be affected (54:52).
That contrast is important. It suggests an uneven transition, where some tasks become easy to automate while other forms of work retain value because they depend on taste, judgment, context, or human preference.
Graphic design shows the split
Altman used graphic designers as an example of how AI can reduce the value of simple commissioned work while increasing the value of human work in other cases.
He pointed to birthday invitations as a task that AI could easily replace. That kind of simple output is vulnerable because it can be generated quickly and cheaply.
But Altman also said that "the price of AI generated art is a zero and the price of human generated graphic art has continued to go up since this has happened" (55:26).
The split is straightforward:
- Routine creative requests can be handled by AI when the goal is a fast, acceptable result.
- Human graphic art can still command value when people care about authorship, originality, or the work of a specific person.
- Job impact depends on the category of work, not just the job title.
This is why the future of work discussion around AI cannot be reduced to one simple outcome. The same technology can erase demand for some tasks while making other human work more valuable.
The warning is about readiness
Altman's comments point to a central tension: AI labs may be moving faster because AI is helping them move faster, while society is still trying to understand the systems already in public use.
His warning that "the world is not prepared" is not presented as a prediction about one narrow industry. It applies to the broader gap between what labs see internally and what the public, workers, and organizations may be ready to handle.
The source article's strongest through-line is acceleration. OpenAI is using AI to support its own research. AGI is described as "pretty close". Superintelligence is described as "not that far off". Software development is already being reframed. Graphic design shows how replacement and rising human value can happen at the same time.
For readers, the key takeaway is that AI progress is no longer only about the next public release. It is also about how AI changes the process of invention itself.