Wannabe stand-up comedians and annoyed prospects, beware: The robotic telephone operators could not discover your deadpan insults amusing sooner or later.
The Info reported that machine studying firm OpenAI may unveil a voice assistant with each audio and visible capabilities that might, in idea, detect sarcasm.
Lastly.
In response to one particular person with data of the brand new tech who spoke with The Info, the thriller assistant may enhance on the automated customer support agent know-how the corporate already provides.
The AI assistant — which might speak to customers and acknowledge objects and pictures — may have many different options, after all, the outlet reported, citing two individuals who have seen it themselves. These options embrace “a greater understanding of picture and audio” and “higher logical reasoning,” per the report.
The know-how might be revealed as quickly as Monday throughout OpenAI’s deliberate livestream asserting updates on their GPT know-how.
“The assistant may theoretically do a variety of issues not potential right this moment, comparable to appearing as a tutor for a pupil engaged on a paper or on math issues, or giving folks details about their environment after they ask for it, like translating indicators or explaining the way to repair automotive troubles,” the report stated.
The brand new multimodal mannequin continues to be liable to AI hallucinations — a phenomenon the place fashions spit out solutions that don’t have any foundation in actuality — an individual aware of it informed The Info.
This new tech may ultimately be built-in into the publicly out there and free model of OpenAI’s fashionable chatbot ChatGPT.
The Info stories the tech will transfer CEO Sam Altman one step nearer to making a extra helpful AI assistant much like the digital Samantha, performed by Scarlett Johansson within the film “Her” — although hopefully, nobody falls in love with it. (That was not sarcasm.)
Representatives for OpenAI didn’t instantly reply to a request for remark.
