OpenAI's head of Codex says the bottleneck to AGI is humanity's inability to type fast enough (www.businessinsider.com)

In a recent discussion on "Lenny's Podcast," Alexander Embiricos, head of OpenAI's Codex, highlighted a surprising bottleneck in the quest for artificial general intelligence (AGI): human typing speed. He argues that the reliance on writing prompts for humans to review AI-generated work significantly limits progress. Embiricos believes that to advance toward AGI, we need to develop AI agents capable of autonomously validating their own output, thereby allowing human users to focus on more complex tasks rather than manual input. Embiricos anticipates that as systems evolve to offload prompt-writing and review responsibilities from humans, we will see a dramatic increase in productivity—a phenomenon he describes as "hockey stick growth." While recognizing that this transition won't happen overnight and that each application will require tailored solutions, he expects early adopters to start experiencing substantial productivity gains as soon as next year. This surge in efficiency is likely to feed back into AI development, accelerating the timeline toward achieving AGI as operational workflows become increasingly automated.