Samsung's new super real digital humans - Samsung Neon

[realpetertdm]

https://www.cnet.com/news/meet-neon-the-artificial-human-startup-funded-by-samsung/

 

https://www.neon.life

Quote

NEON is a computationally created virtual being that looks and behaves like a real human, with the ability to show emotions and intelligence.

 

- Neon -

Neon is developed by Samsung Technology and Advanced Research Labs (STAR Labs) and is funded by Samsung.

 

Quote

The mysterious company, emerging from the Samsung Technology and Advanced Research Labs (aka STAR Labs), debuted late Monday at CES2020 in Las Vegas. It described its technology, also called Neon, as "a computationally created virtual being that looks and behaves like a real human, with the ability to show emotions and intelligence."

Basically, Neon makes avatars that look and act like real people. Neons aren't all-knowing smart assistants, androids, surrogates or copies of real humans, the company said in an FAQ shared with reporters. They can't tell you the weather or how old Abraham Lincoln was when he died. 

 

"Neons are not AI assistants," the company said. "Neons are more like us, an independent but virtual living being, who can show emotions and learn from experiences. Unlike AI assistants, Neons do not know it all, and they are not an interface to the internet to ask for weather updates or to play your favorite music."

Instead, they're designed to have conversations and behave like real humans. They form memories and learn new skills but don't have a physical embodiment, at least not now. Neons can help with "goal-oriented tasks or can be personalized to assist in tasks that require human touch." They can act as teachers, financial advisers, health care providers, concierges, actors, spokespeople or TV anchors. 

While they can borrow traits of real people and have a similar look and voice, they can't be exact copies of existing humans, Neon said. And each Neon is unique, with its own personality. 

 

"There are millions of species on our planet, and we hope to add one more," Pranav Mistry, Neon CEO and head of STAR Labs, said in a press release. "Neons will be our friends, collaborators and companions, continually learning, evolving and forming memories from their interactions."

Quote

While Neon is funded by Samsung and is part of Samsung's STAR Labs, its avatars are not a Samsung product. It also isn't related to Bixby, Samsung's digital assistant that first appeared in 2017's Galaxy S8. STAR Labs describes itself as a "fully independent future factory with a mission to bring science fiction to reality."

Quote

The company said it will preview the technology at something it called Neonworld 2020 later this year. It didn't say when or where that would take place. 

(...)

The company plans to beta-launch Neon later this year with certain partners around the world. It said that people will be able to license or subscribe to Neons for expert help -- asking them for financial advice, for example. It expects them to one day appear in movies and on the news. 

Very interesting. If what they say is true then this could open a whole new chapter in the history of artificial intelligence. So far a lot of the information is hidden, but still intriguing nontheless. 

[mr mans]

I doubt we are anywhere near the level of having even an empty conversation with a computer 

[williamcll]

If it's not streaming on youtube like something out of an anime I'm not interested.

[Bombastinator]

1 minute ago, vukos said:

I doubt we are anywhere near the level of having even an empty conversation with a computer 

The touring AI test invokes just that.  The theory as I have been given to understand it goes that to make an AI that is effective one makes a program that cannot be recognized as not being human.

My memory is that the most successful ones so far have emulated internet trolls that spew abuse at people.  I’m am not sure this is an actual proper use of the test but is more of an attempt to make an end run around it.

[Bombastinator]

29 minutes ago, huilun02 said:

So its basically an actor that has to be trained to give a canned response or targeted result

Seem they are being very careful with their wording to try and sell it as AI but without making any promises

Perhaps.  Such things as I understand them are referred to as “mechanical Turks” and are not actually AI.

[Brooksie359]

This seems interesting but at the same time problematic. It sounds like they are trying to get computers to even do the few jobs that require the "human touch". Those are the jobs that are supposed to still be viable even after computers replace people for all jobs that do not require a human touch. Now if these things can basically function just like people but better then I could see a future where literally all jobs are lost to AI. 

[Bombastinator]

14 minutes ago, Brooksie359 said:

This seems interesting but at the same time problematic. It sounds like they are trying to get computers to even do the few jobs that require the "human touch". Those are the jobs that are supposed to still be viable even after computers replace people for all jobs that do not require a human touch. Now if these things can basically function just like people but better then I could see a future where literally all jobs are lost to AI. 

I’d say it’s only problematic if it works.  I seriously doubt it works very well.  It is an issue with our coming future as a species though.  

[realpetertdm]

8 hours ago, Brooksie359 said:

This seems interesting but at the same time problematic. It sounds like they are trying to get computers to even do the few jobs that require the "human touch". Those are the jobs that are supposed to still be viable even after computers replace people for all jobs that do not require a human touch. Now if these things can basically function just like people but better then I could see a future where literally all jobs are lost to AI. 

If these things can actually think/imagine/feel like a human then it could very well be a problem. AI and robots are already taking over many jobs. We probably need some regulations or laws that deal with this issue going forward.

 

Also if these AIs can think like humans then can they develop AI too? Will it get to a point where robots make robots?

[Xee]

Watching LTT CES Neon coverage I was smiling more and more thinking - "yeah I heard that before - in some sci-fi movie or book". The best part was "forgetting every interaction as soon as it is over".

Based on that I got two excellent avatar ideas:

[TechyBen]

On 1/7/2020 at 2:56 PM, Bombastinator said:

The touring AI test invokes just that.  The theory as I have been given to understand it goes that to make an AI that is effective one makes a program that cannot be recognized as not being human.

My memory is that the most successful ones so far have emulated internet trolls that spew abuse at people.  I’m am not sure this is an actual proper use of the test but is more of an attempt to make an end run around it.

Yeah, but also, there are limits to the real touring test/competition. IRL you could just ask "what time is it", and an AI not connected to some sort of data/simulation won't know if it's in UK, USA or Japan, or even have access to the computer clock in the OS/hardware it's running. If the simulation/AI has access to *external* data, then that's different ("what did the Queen do last week?").