OpenAI’s latest innovation, GPT-5, touted as “PhD level” intelligence, has sparked excitement but also confusion among users. At a recent launch event, CEO Sam Altman illustrated GPT-5 as “a PhD-level expert in your pocket,” surpassing previous versions. However, users discovered critical errors, like misidentifying elements in words and misleading information regarding US states. For instance, it mistakenly noted that “blueberry” has three Bs and miscalculated the Rs in “Northern Territory,” claiming there were three instead of five. OpenAI’s “real-time router” aims to enhance response accuracy by selecting models based on user queries. Although GPT-5 represents a leap toward artificial general intelligence (AGI), it still falls short, as noted by analysts who found it sometimes generates hallucinated answers instead of factual ones. Despite these glitches, GPT-5 is now accessible to OpenAI’s vast user base of 700 million, marking a significant advancement in AI technology.
Source link
OpenAI Launches ChatGPT-5: Advanced AI Struggles with Basic Spelling and Geography Despite ‘PhD-Level’ Claims | Australia News
Share
Read more