Samsung’s STAR Labs led by Pranav Mistry has finally lifted the veils off their mysterious NEON project that has been hyped on social media for a while now. In a nutshell, NEON is an Artificial Human that looks and behaves like a real human and have the ability to show emotions and intelligence. While that might not sound as groundbreaking since we’ve already seen various image and video manipulation techniques such as deep fakes or CG, it’s categorically different.
So, what is NEON? Why is it different?
NEON, as mentioned above, is an Artificial Human that looks surreally real and emote and converse as you’d expect a human to do. NEONs are powered by what STAR Labs calls CORE R3. CORE R3 stands for Reality, Realtime and Responsive, and is what gives NEONs the power to computationally create a lifelike reality that is “beyond normal perception to distinguish”. If you take a look at the demos showcased, you won’t find a reason to disagree.
It’s important to know that NEONs aren’t ubiquitous know-it-alls like the voice assistants we’ve become so used to. Unlike Google Assistant, Amazon’s Alexa, or Apple’s Siri, NEON is an independent virtual being that learns from experience and builds memories. But unlike the voice assistants mentioned above, NEON can communicate with human effect, by understanding and sympathizing just like a real human.
Flying to CES tomorrow, and the code is finally working 🙂 Ready to demo CORE R3. It can now autonomously create new expressions, new movements, new dialog (even in Hindi), completely different from the original captured data.
STAR Labs is still working on SPECTRA that will complement Core R3 with the spectrum of Intelligence, Learning, Emotions, and Memory, making NEON more immersive.
Also, unlike deep fake or other video manipulation techniques where existing images or videos are manipulated, NEONs (with the help of CORE R3) create unique behaviors and interactions in real-time. Or as STAR Labs states, NEONs create new realities.
Hence, even though a NEON can borrow similarities or be inspired from a real person’s look or voice, STAR Labs says every NEON has its own unique personality. And yes, NEONs can be created fully from scratch too.
Why NEON? Are they aimed at replacing humans?
At their booth at CES 2020, STAR Labs showcased NEON in several personas, namely: a yoga instructor, a banker, a K-pop star, a news anchor, and a fashion model amongst others. While this isn’t happening anytime soon, NEONs are eventually expected to make inroads into job applications pertaining to the professions mentioned above.
NEONs can assist with goal-oriented tasks and can be customized to assists in tasks that require a human touch. The company says it could serve as an individualized teacher, a personal financial advisor, a healthcare provider, or a concierge. But what’s more ambitious is that NEONs can also be an actor, a spokesperson, or a TV anchor (like seen in the demo videos).
What about security and privacy?
Since NEONs can be modeled after real human beings, it’s natural to think about the unethical usage that will crop up with it. While there’s no clarity on this aspect, STAR Labs claims that the CORE R3 architecture was designed around the principles of privacy and trust.
“CORE R3 ensures the integrity of data with state-of-the-art security protocols. No one except you and your NEON can ever have access to your interactions with NEON. NEONs will never share your private data without your permission.”
When will the NEON launch?
The above-mentioned applications of the NEON are still quite far away from actually being implemented. Although, NEON is expected to beta launch in the real world with a few partners later this year, in 2020. No information regarding the pricing has been announced by the company yet.