In recent years, artificial intelligence (AI) has seen significant growth across all sectors. The rapid adoption of AI technology has become so prevalent that any tech not utilizing AI may seem less valuable. As AI systems become more advanced, the need to measure their intelligence has become more critical. OpenAI has addressed this by developing an internal scale to track the progress of its large-language models toward artificial general intelligence (AGI). According to a recent report, this scale is designed to evaluate the intelligence of OpenAI’s tools.
A Bloomberg report revealed that OpenAI’s ChatGPT is currently at Level 1 on this scale. OpenAI created this level-based scale to measure progress, with each level defined by specific criteria. These levels are compared to human abilities in solving tasks.
OpenAI measures AI intelligence
The AI intelligence framework outlines five levels of sophistication. OpenAI claims that while ChatGPT is currently at Level 1, it is quickly approaching Level 2. Level 2 AI corresponds to the problem-solving abilities of a PhD holder, while Level 3 AI can autonomously perform tasks on a user’s behalf. Level 4 AI demonstrates creative capabilities, generating novel innovations.
Level 5 AI, the highest level, can replicate the collective efforts of an entire organization, effectively achieving AGI. OpenAI has defined AGI as “a highly autonomous system surpassing humans in most economically valuable tasks.”
This level system indicates that OpenAI aims to achieve AGI-level intelligence with its tools. A few months ago, OpenAI CEO Sam Altman made headlines by stating that he wants to build AGI, no matter the cost.
OpenAI also revealed that the GPT-4 model exhibits new skills demonstrating human-like reasoning, according to the Bloomberg report. This internal scale follows hints from the OpenAI CEO about the company’s next large language model, GPT-5. According to Altman, GPT-5 will be “a significant leap forward.”
OpenAI ChatGPT 5
In a report by The Decoder, Altman spoke at the Aspen Ideas Festival, expressing optimism about GPT-5. However, he also noted that there is still much work to be done on the new model. Despite this, he expressed confidence in its potential, stating that GPT-5 aims to address many of the errors seen in GPT-4.
While details about ChatGPT-5 remain unclear, Altman’s remarks suggest that the model is still in the early stages of development. The anticipation surrounding GPT-5 follows the impressive debut of GPT-4, which showcased its mathematical prowess and ability to engage in emotionally intelligent conversations. OpenAI introduced GPT-4 during its Spring Update event in May this year.
 has seen significant growth across all sectors. The rapid adoption of AI technology has become so prevalent that any tech not utilizing AI may seem less valuable. As AI systems become more advanced, the need to measure their intelligence has become more critical. OpenAI has addressed this by developing an internal scale to track the progress of its large-language models toward artificial general intelligence (AGI). According to a recent report, this scale is designed to evaluate the intelligence of OpenAI's tools.){kind=link}