OpenAI recently unveiled a new system consisting of five levels to monitor its advancement towards developing artificial intelligence software surpassing human capabilities. The startup, known for its cutting-edge AI technology, disclosed this classification system to its employees, designed to enhance transparency regarding its approach to AI safety and evolution.
At the core of the classification system are five distinct levels, each signifying a different stage of AI development. The levels range from the current conversational AI solutions available today, categorized as Level 1, to the envisioned AI systems capable of replicating organizational functions, identified as Level 5.
During a company-wide meeting, OpenAI officials indicated that they believe the organization is currently at Level 1 but is on the verge of progressing to Level 2, labeled as Reasoners. This tier refers to AI systems that can perform basic problem-solving tasks at the level of a highly-educated individual without external aids.
Moreover, OpenAI showcased a demo of a research project involving its GPT-4 AI model, highlighting new capabilities resembling human-like reasoning. While not currently accessible to the public, this demonstration underscored the continuous efforts within OpenAI to push the boundaries of AI capabilities.
The ultimate goal for OpenAI remains the development of artificial general intelligence (AGI), encompassing AI systems that excel at diverse tasks beyond human abilities. Through its tiered system, the company aims to provide clarity on its roadmap towards achieving AGI, aligning with CEO Sam Altman’s vision of potentially reaching AGI within this decade.
OpenAI’s tiered approach echoes similar frameworks proposed by industry peers, emphasizing the pursuit of increasingly advanced AI capabilities. As the organization seeks input from stakeholders and refines its classification system, the journey towards AGI continues to unfold, shaping the future landscape of artificial intelligence.
