OpenAI’s “magic” GPT-4o update is an iterative release that may herald less AI hype

OpenAI CEO promised “magic” at this week’s announcement, but instead delivered something better: an iterative update to ChatGPT, including whizzy new voice features. So does OpenAI’s GPT-4o update mark a new, humbler era for generative AI?

The first signs appeared last week when OpenAI said it would be demonstrating new features and updates to its AI models and ChatGPT. At the time, CEO Sam Altman stressed the demo wouldn’t include GPT-5 or a much-rumoured search tool to take on Google, but did say what would be on show “feels like magic to me”.

In a 26-minute-long video of what the AI company is calling a “Spring Update“, Chief Technology Officer Mira Murati began by saying: “Today I’m going to talk about three things — that’s it.”

OpenAI unveiled a desktop version, a refreshed web UI, and an update to its model, bringing it to GPT-4o (that’s a letter “o” after the 4, to be clear). We explain exactly what’s new in the updated in our guide, what is GPT-4o?

Perhaps the biggest news is that the new model will be available to free users rather than just paid subscribers.

End of the AI hype?

Generative AI is overwhelmingly hyped, with huge leaps and bounds in terms of progress with each passing month. Or at least, that’s how things have been.

With this more modest release, OpenAI’s work sits within the cadence of the existing tech industry where it’s one big launch per year. But that doesn’t stop the likes of Apple, Microsoft and Google creating slick demos for iterative updates, as we saw from OpenAI this week.

Indeed, OpenAI’s Spring Update came a day before Google’s I/O Developer Conference.

It remains to be seen whether OpenAI’s iterative updates and demos of not-quite-ready features are a sign of market maturity or a small dip before returning to full-on hype — after all, Altman still promised “magic”.

Desktop app and new UI

The AI industry likes to big up the technology with talk of existential threats; Altman himself has claimed the company is “a little bit scared” of what it’s building. But instead of world-ending AI, Murati unveiled a desktop app to help users better integrate ChatGPT into their workflows, by opening the chatbot with a keyboard shortcut.

This is similar to Microsoft’s Copilot key and Logitech’s AI shortcuts for its keyboards and mice.

Initially, the ChatGPT app will only be available for macOS, with a Windows version coming later this year — despite Microsoft’s hefty investment in OpenAI.

OpenAI also unveiled a simplified look for the web version of ChatGPT, to help users “not focus on the UI at all, but focus on the collaboration with ChatGPT,” Murati said.

Change in direction?

Will this modest set of demos — a new UI, app and voice chat — help reduce the hype? Probably not, but a more informed discussion of what these technologies are capable of and limited to is certainly welcome.