June 22, 2024 – In an interview released this Thursday by Dartmouth Engineering School, Mira Murati, the Chief Technology Officer of OpenAI, likened the leap from GPT-4 to GPT-5 to the growth from a high school student to a PhD holder.
“If you look at the evolution trajectory of GPT, systems like GPT-3 might have been at the intelligence level of a toddler, while GPT-4 is more like a smart high school student,” Murati said. “In the next few years, we expect to achieve a PhD-level intelligence for specific tasks. Things are rapidly changing and improving.”
Murati emphasized that this “PhD-level” intelligence would only apply to certain tasks. “These systems have already reached human-level performance in specific tasks, although in many others, they are not yet up to that level,” she clarified.
When asked about a specific timeline for GPT-5’s release, Murati nodded affirmatively and indicated that it would be launched in about a year and a half. This suggests that GPT-5 might make its debut towards the end of 2025 or early 2026.
Echoing similar sentiments, other tech leaders have also expressed their views on the rapid advancement of AI. Microsoft’s Chief Technology Officer, Kevin Scott, has stated that the next generation of AI systems will possess enhanced memory and reasoning abilities, potentially even passing a PhD examination. Likewise, Alibaba Chairman Joe Tsai mentioned, “It’s amazing that in just three or four years, we’ve managed to bring large language models to a level comparable to humans in terms of knowledge and certain mathematical calculations, and even to the level of a PhD student in some aspects. The speed of this progress is both frightening and incredible.”