Humane's AI Pin may have made a rough start, but we expect OpenAI's latest model to help revive its ailing AI hardware business The company said this week that Humane's AI Pin is working with OpenAI's new ChatGPT-4o model In the X post, Humane automatically pushed the compatibility of ChatGPT-4o and said that users don't need to do anything special to get updates

"The CosmOS architecture allows us to quickly adopt the latest and best AI technology," Humane says of Pin's software

OpenAI recently announced the ChatGPT-4o model, which the company said is the most capable to date In addition to text, the model can accept any combination of voice, image, and video, as well as providing the same multimedia response OpenAI is also faster than the previous GPT-4 version and is much better at understanding and delivering both audio and visual data than the previous model It was a great opportunity for me to be a part of the team

Humane itself stopped saying exactly how the ChatGPT-4o integration would benefit Humane AI pins, but product design lead George Kedenburg said, based on internal testing,

"So far, many of the major improvements," Kedenburg writes in X"[A] 14% of the latency has been reduced" There were fewer and fewer bad answers, with a decline, 28% short answers and 33% over the figure Everything just feels (predictably) smarter and more accurate

This is good news for Humane, a company that has suffered its fair share of bad news since launching Pin earlier this year

The device had a lot of hype up until its launch, but once users got the device, there was a bug, there was no context, and sometimes flatly, as soon as reviewers got the pin, they immediately panned the pin because of poor performance, software bugs, etc It was a great experience

pin is questioning whether AI-based hardware can or should be successful The model they rely on obviously works well and helps people get the experience they want, but at least so far, the AI hardware has not met expectations