The race to build artificial general intelligence continues to intensify with large tech companies investing billions in smarter AI models and powerful computing systems. However, a senior OpenAI leader says the biggest problem is no longer technology; the real issue is human speed.
Alexander Embiricos, who leads the Codex project at OpenAI, recently shared this view during a public discussion. He explained that modern AI systems already perform complex tasks. These systems can also write computer programs, debug them, and manage big workloads. The reason for slow progress is that human interaction with AI is not fast enough.
Writing prompts and checking AI results are time-consuming. The delay limits how much work AI can complete. Embiricos called this a human speed limit. While AI keeps improving quickly, human input does not scale at the same pace.
He said current workflows depend too much on manual steps. Humans still guide AI at every stage. Every output needs review. Every task needs approval. This constant oversight reduces the real benefit of advanced systems. Even strong AI tools lose value when people must monitor every action.
This issue highlights the growing limits of AI interaction. In the past, experts focused on computing power and model size. Today, those areas move fast. Human involvement now creates the biggest delay. According to Embiricos, this problem affects many industries using AI tools.
He believes the next phase of growth needs a major shift. AI systems must work reliably on their own. They should require less checking and fewer corrections. When systems become dependable by default, productivity can rise sharply.
Embriicos described this growth as gradual at first, then sudden. Early users of better automation may see clear gains soon. Larger companies may follow as trust in AI systems increases. This change could drive strong AI productivity growth across sectors.
However, he also warned against simple solutions. Every use case differs. Fully autonomous AI systems need careful design. Some tasks require more control. Others allow more freedom. The path to AGI will involve many steps, not one breakthrough.
Moreover, this change affects the future of work. By the time AI automation reaches its peak, the human roles might have changed. The workers might be engaged only in planning, decision-making, and creative work. Routine typing and checking may become less common.
For years, the AGI race focused on faster chips and bigger models. Embiricos adds a new perspective. Human typing speed now plays a central role. Progress depends on better system design and smoother human-machine interaction.
The message is clear. Smarter AI alone is not enough. Reducing human friction may decide how fast AGI becomes real.
Also Read – OpenAI Drops GPT-5.2 to Outpace Google Gemini in AI Power Race