OpenAI’s AI models need to crack these 5 stages to achieve human-like intelligence

While there is no consensus on what AGI actually means, OpenAI has earlier said it is when a highly autonomous system surpasses humans in most economically valuable tasks.

Achieving AGI will reportedly take a massive amount of computing power that would cost billions of dollars. OpenAI CEO Sam Altman, in October 2023, said that the emergence of AGI will take “five years, give or take.”

OpenAI’s road to AGI

OpenAI’s tiered approach to AGI will reportedly involve classifying its AI models and their subsequent iterations based on unlocking new capabilities.

– Level 1 refers to the AI capabilities seen in present-day chatbots.

– Level 2 is for models that can provide solutions to basic problems with the intelligence of a PhD holder.

– Level 3 involves developing AI models that can take actions on a user’s behalf.

– Level 4 is meant for AI models that can come up with entirely new innovations.

– Level 5 is when AI models can carry out work that was earlier done by multiple human organisations. It is considered to be the last stage in achieving AGI.

While ChatGPT is currently at Level 1, OpenAI has reportedly said with GPT-4o, it is close to levelling up to the next one. However, the finer details of OpenAI’s classification methods are not yet clear.

Though OpenAI is keen to build AGI before anyone else, the company has also said that it is willing to offer its support to any “value-aligned, safety-conscious project” that comes close to developing the revolutionary AI technology. Meanwhile, concerns about OpenAI’s safety priorities have grown since the exit of the company’s cofounder Ilya Sutskever as well as key OpenAI researchers.