Non Paywalled: https://archive.is/ttaPF
It may not have been OpenAI’s intention, but what the launch of GPT-5 makes clear is that the nature of the AI race has changed.
Instead of merely building shiny bigger models, says Sayash Kapoor, a researcher at Princeton University, AI companies are “slowly coming to terms with the fact that they are building infrastructure for products”.
I think that it would be good if there's a plateau in the current capabilities, because:
I've actually thought that for a long time: #892431
The AI race won't be who has the best model, because it seems like the gains to capabilities are already experiencing diminishing marginal returns.
The AI race will be won on new features, or perhaps clever ways to cost cut the existing models.