OpenAI CEO Clarifies: AGI Development Not Achieved, Urges Calmness

OpenAI's CEO, Sam Altman, has once again addressed the public, urging calm regarding the topic of artificial general intelligence (AGI). In a social media post on X, he wrote: "The hype on X is getting out of hand again. We will not be deploying AGI next month, nor are we building it." He further added: "Please take a step back and lower your expectations by at least 100 times!"

These comments seem to contradict some of Altman's previous statements. In his blog post, Altman had mentioned that OpenAI knows how to build AGI and predicted that this year humans would collaborate with AI agents. He even hinted that superintelligence could emerge within "a few thousand days". On January 5th, he wrote in his blog: "We now have confidence that we know how to build AGI in the traditional sense. We believe that by 2025, we may see the first AI agents 'joining the workforce', which will fundamentally change company outputs."

However, Altman's remarks have stirred up the AI community and raised doubts among experts. Humayun Sheikh, CEO of Fetch.ai and chairman of the ASI Alliance, commented: "When considering Altman's views on AGI-level AI agents, it is crucial to understand how the definition of AGI is evolving. AGI has not yet reached true consciousness, and I believe it won't for quite some time."

It's worth noting that this isn't the first time Altman has walked back his own hype. After heavily promoting the capabilities of GPT-4o, he later retracted his claims, calling ChatGPT "at best somewhat embarrassing".

Altman himself has reflected on how people debate the potential impacts of new AI models, noting that due to unmet expectations, individuals often end up demanding more features from these models. In one tweet, while acknowledging disappointing promises, he indicated that OpenAI is working on some "cool stuff", but cautioned against expecting it to pass the Turing test anytime soon. In a prior X post, he confirmed that the OpenAI o3 mini (a less powerful version of the full o3 inference model) would be released in a few weeks.

To date, OpenAI has not responded to requests for comments on specific functionalities or release timelines for the o3 model.