A leader in artificial intelligenceOpenAI, recently released a newAGIThe AI roadmap divides AI capabilities into five levels. Once released, this five-level roadmap has attracted widespread attention and heated discussions in the industry. However, it has also been criticized and questioned by some professionals and netizens.
According to OpenAI's plan, AI is currently at the L1 stage, which is the chatbot level. The next four levels are: L2 reasoner (human-level problem solving), L3 agent (system that can take action), L4 innovator (artificial intelligence that can assist in invention), and L5 organizer (artificial intelligence that can complete organizational work). Through this structured scale, OpenAI hopes to provide a clear framework for the progress of AI and ensure that every step towards AGI is measurable and responsible.
Source Note: The image is generated by AI, and the image is authorized by Midjourney
However, this seemingly perfect "five-step strategy" has been questioned by many people, including Musk. They believe that OpenAI's roadmap is vague in naming and definition, and the five-level hierarchy is "logically confusing." Some even think that this may just be a pie in the sky for investors, an empty marketing trick.
Some criticisms have focused on OpenAI's definition of "superintelligence." The concept of artificial superintelligence (ASI) is controversial and lacks a clear and widely accepted definition. Defining a certain level as ASI could lead to misunderstandings and unrealistic expectations or concerns about the capabilities and potential impacts of AI.
In addition to the confusion in concepts and definitions, the rationality of this five-level order has also been questioned. Some people point out that this order is unreasonable according to the development level of human capabilities. For example, some independently operated agents are already performing L3-level tasks, while some people with doctoral degrees may not be able to complete L2-level tasks. This makes people wonder: Is this five-level order really tenable?
Is OpenAI's five-level AGI roadmap a major breakthrough in the field of artificial intelligence or just an empty marketing hype? This remains to be tested by time and practice. But it is undeniable that it has triggered people's thinking and discussion on the future development of artificial intelligence, which is meaningful in itself.
In this era of rapid development of artificial intelligence, we need more rational voices to guide and promote the healthy and orderly development of this field. As for OpenAI's roadmap, we might as well keep an eye on it and wait and see.