OpenAI's new product could include sarcasm detection

The Information reported that machine learning company OpenAI could unveil a voice assistant with both audio and visual capabilities that could, in theory, detect sarcasm.

Finally.

According to one person with knowledge of the new tech who spoke with The Information, the mystery assistant could improve on the automated customer service agent technology the company already offers.

The AI assistant — which can talk to users and recognize objects and images — could have many other features, of course, the outlet reported, citing two people who have seen it themselves. These features include "a better understanding of image and audio" and "better logical reasoning," per the report.

The technology could be revealed as soon as Monday during OpenAI's planned livestream announcing updates on their GPT technology.

"The assistant could theoretically do a range of things not possible today, such as acting as a tutor for a student working on a paper or on math problems, or giving people information about their surroundings when they ask for it, like translating signs or explaining how to fix car troubles," the report said.

The new multimodal model is still prone to AI hallucinations — a phenomenon where models spit out answers that have no basis in reality — a person familiar with it told The Information.