GPT-5
A frontier large language model referenced as the current state-of-the-art at the time of recording, scoring 64% on APEX.
As defined in Brendan Foody
Where it comes up · 4
- AI Experts Debate: AI Job Loss, The End of Privacy & Beginning of AI Warfare w/ Mo, Salim & Dave 1762 claims
- Brendan Foody1 claim
- Eliezer Yudkowsky — Why AI will kill us, aligning LLMs, nature of intelligence, SciFi, & rationality1 claim
- Satya Nadella — How Microsoft is preparing for AGI1 claim
Most important claims · 5 of 5Search all claims for “GPT-5” →
- Scaffolding—the ability to layer AI capabilities on top of foundational models to make them useful for specific domains—is more important than the next generation of the foundation model itself, so entrepreneurs should not wait for GPT-5 but should use current models and optimize for their use case.AI Experts Debate: AI Job Loss, The End of Privacy & Beginning of AI Warfare w/ Mo, Salim & Dave 176
- The rate of model improvement at economically valuable tasks is roughly 25-30 percent per year; GPT-5 now scores 64 percent on APEX, up from GPT-4o a year ago, though sustaining that rate would be challenging.Brendan Foody
- GPT-5 is expected to launch in July 2025 with PhD-level reasoning ability and self-coding capability, representing the expected progression along the scaling law curve rather than a discontinuous leap.AI Experts Debate: AI Job Loss, The End of Privacy & Beginning of AI Warfare w/ Mo, Salim & Dave 176
- A moratorium on further AI training runs should be enacted now rather than waiting for GPT-5, because algorithms keep improving and getting more efficient per FLOP, so starting a freeze at a higher capability level leaves much less lifeline before reaching dangerous territory, and because earlier systems could become entrenched making a later stop harder both politically and technically.Eliezer Yudkowsky — Why AI will kill us, aligning LLMs, nature of intelligence, SciFi, & rationality
- Microsoft's Fairwater 2 datacenter represents a roughly 10x increase in training capacity over what GPT-5 was trained with, achieved by 10x-ing training capacity every 18-24 months, with network optics in a single building nearly equal to all of Azure two and a half years ago.Satya Nadella — How Microsoft is preparing for AGI