|
That's what 22-year-old Stanford University computer science
student Bryan Chiang was wondering earlier this year.
So, he grabbed an augmented reality eyepiece and his laptop and
recruited a few friends to code what he calls RizzGPT.
The eyepiece - a monocle designed by Brilliant Labs that is
open-sourced so its firmware can be experimented with - features
a camera, a microphone and an internal projector screen where
words are displayed in front of the user's eye.
When someone talks to the user, RizzGPT monitors the
conversation through the microphone, transforms it to text, and
sends it via WiFi to OpenAI's artificial intelligence chatbot
ChatGPT to generate a response. That response then appears after
a short delay on the small monocle screen.
"RizzGPT basically uses AI to provide you charisma on demand,
and so it listens to your current ongoing conversation, and it
tells you exactly what to say next," said Chiang.
In a demonstration, Reuters asked Chiang: "What do you see as
your biggest weakness?"
"I believe my biggest weakness is that I can be too hard on
myself sometimes. I'm always striving to do my best and
sometimes I can burn myself out," Chiang read from the monocle
after about five seconds.
The delay and the response is not yet very natural - or
charismatic. But it is strictly a prototype, intended to show
what may be possible with the technology, Chiang said.
"It's been a while since how we interact computers has changed,"
he said. "You're seeing the convergence of 5G connectivity, AR
glasses, the hardware, the intelligence coming together to
basically create a new way of interacting with these systems, a
new operating system in which it's much more natural."
The goal was not to replace natural human conversation entirely,
he said.
"It's merely meant as this sort of assistive aid to help you
think about things that you might have forgotten... I think in
that role it could be incredibly helpful for people who struggle
with social anxiety and have difficulties, you know, talking to
others."
(Reporting by Nathan Frandino, Editing by Rosalba O'Brien and
Cynthia Osterman)
[© 2023 Thomson Reuters. All rights
reserved.]
Copyright 2022 Reuters. All rights reserved. This material may
not be published, broadcast, rewritten or redistributed.
Thompson Reuters is solely responsible for this content.
|
|
