OpenAI's head of Codex says the bottleneck to AGI is humanity's inability to type fast enough

The race for artificial general intelligence (AGI) is heating up, and a surprising factor has emerged as a potential roadblock: human typing speed. Alexander Embiricos, the head of product development for Codex, OpenAI's coding assistant, shared his insights on this topic during a recent episode of 'Lenny's Podcast'. Embircious pointed out that the "current underappreciated limiting factor" in the pursuit of AGI is not just the technology itself, but rather the speed at which humans can type and multitask when providing writing prompts. He emphasized that while AI systems can monitor and assist in various tasks, they still rely heavily on human input to validate their outputs. This dependency creates a bottleneck in efficiency. To overcome this challenge, Embiricos believes it is crucial to redesign systems that minimize the burden of prompt writing and validation on humans. He envisions a future where AI agents can operate with greater autonomy, significantly enhancing productivity. "If we can rebuild systems to let the agent be default useful, we’ll start unlocking hockey sticks," he stated, referring to the dramatic growth curve in productivity that could follow. Although there is no straightforward solution to creating fully automated workflows, Embiricos is optimistic about upcoming progress. He predicts that early adopters will begin to experience significant productivity boosts as soon as next year. This trend is expected to accelerate as larger companies integrate these advancements over the following years. Embircious also speculated that the period between these productivity gains among early adopters and the full automation of processes would be critical for achieving AGI. He noted, "That hockey-sticking will be flowing back into the AI labs, and that's when we'll basically be at the AGI," highlighting the interconnected nature of human productivity and artificial intelligence advancement.