OpenAI’s always-on AI device is plagued by various issues

OpenAI and former Apple designer Jony Ive announced work on a mysterious new AI device last year, and we’ve heard a few details since then. Now, a new report has revealed specific issues with this upcoming device.

The Financial Times (FT) reports that OpenAI has encountered a variety of obstacles as it develops the new AI gadget. Furthermore, the outlet’s sources say these issues could delay the device’s release. However, one source said these problems were typical of the product development process.

These major issues include deciding on the assistant’s personality and voice, privacy issues, and “budgeting for the computing power needed to run OpenAI’s models on a mass consumer device.”

A source told the outlet that computing power is still an issue for OpenAI at large:

Amazon has the compute for an Alexa, so does Google, but OpenAI is struggling to get enough compute for ChatGPT, let alone an AI device — they need to fix that first.

It’s possible to run some AI models locally, and we’ve even seen OpenAI release its first AI model that can (barely) run on-device. But even this distilled model requires a powerful Snapdragon device with a ton of RAM. So a cloud-powered device should drive down hardware costs and offer a more powerful experience. However, the last thing you want is a cloud-first device that struggles to actually connect to the cloud.

As for the AI experience itself, the team was apparently trying to ensure that the device’s assistant didn’t talk too much or take too long to finish a conversation. A source also described the assistant’s approach as akin to “a friend who’s a computer who isn’t your weird AI girlfriend… like Siri but better.” Another source also said that the assistant’s personality was a tough balancing act, explaining that it couldn’t be too sycophantic or too direct.

This palm-sized gadget supposedly lacks a screen, but people can apparently use the device via a microphone, camera, and speaker. It’s also believed that the device is meant to sit on your desk or table, but can be carried around. The Financial Times adds that the device is meant to improve on current smart speakers by offering a wider variety of functions.

Another source echoed a previous report and stated that the OpenAI device would be always-on rather than requiring a wake word. They added that the gadget’s sensors would help collect info throughout the day to build the assistant’s “memory.” The FT report didn’t clarify the privacy issues, but I imagine that an always-on experience would indeed be a privacy nightmare.