Artificial Intelligence thread

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member

sunnymaxi

Captain
Registered Member
Please, Log in or Register to view URLs content!

World’s first AI child unveiled in Beijing by award-winning Chinese scientist who left US for China​

  • The AI entity has been unveiled at an exhibition by the Beijing Institute for General Artificial Intelligence
  • Called Tong Tong, or Little Girl, she can assign herself tasks, learn autonomously and explore her environment
Science fiction has now become reality with the creation of Tong Tong, a girl who is the world’s first virtual
Please, Log in or Register to view URLs content!
(AI) entity, developed by Chinese
Please, Log in or Register to view URLs content!
.

Tong Tong, or Little Girl in English, the
Please, Log in or Register to view URLs content!
entity was unveiled at the Frontiers of General Artificial Intelligence Technology Exhibition held in Beijing on January 28-29, under the auspices of the Beijing Institute for General Artificial Intelligence (BIGAI).

There, exhibition visitors could interact with Tong Tong. When programmed to like things tidy, she would fix a crooked picture frame on her own. If the frame was too high for her to reach, she would find a stool so she could straighten it without any help from a human. Also, if someone spilled milk, she would find a towel and clean it up by herself, showing her ability to interpret human intentions

Unlike popular
Please, Log in or Register to view URLs content!
in AI, Tong Tong can independently assign tasks to herself, ranging from exploring her environment to tidying rooms and cleaning stains.
She is capable of autonomous learning and “has her own joy, anger and sorrow”, according to a video posted by BIGAI.

“Tong Tong possesses a mind and strives to understand the common sense taught by humans. She discerns right from wrong, expresses her attitudes in various situations, and has the power to shape the future,” the video says.

In terms of general artificial intelligence standards and testing tasks, Tong Tong displays behaviour and abilities akin to those of a three or four-year-old child. Through exploration and human interaction, she can continually improve her skills, knowledge and values.

A key aspect of general intelligence, as one researcher explained, is having physical and social common sense akin to humans. Driven by its own values, an AI entity should not only be capable of completing an infinite array of tasks, but should also define new ones autonomously.

“To advance towards general artificial intelligence, we must create entities that can comprehend the real world and possess a wide range of skills,” BIGAI director Zhu Songchun said.
 

Overbom

Brigadier
Registered Member
I am genuinely shocked and awed. The world is changing in front of our own eyes.

OpenAI is the undisputed leader in AI.

Below are some twitter links to some of the videos. However as twitter is doing its usual garbage compression on uploaded videos, you might want to visit OpenAI's website to watch the videos on the highest quality.
Please, Log in or Register to view URLs content!







Please, Log in or Register to view URLs content!
 
Last edited:

gadgetcool5

Senior Member
Registered Member
The bar is definitely going up for China if it wants to stay in the game. Right now its generative Chatbots are not even available overseas and you need a Chinese phone number to even use Erniebot, for instance.
 

Bellum_Romanum

Brigadier
Registered Member
The bar is definitely going up for China if it wants to stay in the game. Right now its generative Chatbots are not even available overseas and you need a Chinese phone number to even use Erniebot, for instance.
China is in the game along with the U.S. which obviously leads the field in the realm of L.L.M. A.I. but you make it seem like this is the only area in A.I. and nothing else. Not to mention that besides U.S. and China no other country comes close to them in terms of A.I. prowess judging from the products they produce to the papers published on A.I. Please indulge us who else has the advantage over China in this regard @gadgetcool5
 

Overbom

Brigadier
Registered Member
I am genuinely shocked and awed. The world is changing in front of our own eyes.

OpenAI is the undisputed leader in AI.

Below are some twitter links to some of the videos. However as twitter is doing its usual garbage compression on uploaded videos, you might want to visit OpenAI's website to watch the videos on the highest quality.
Please, Log in or Register to view URLs content!







Please, Log in or Register to view URLs content!
To understand the implications of this, in order to create realistic video you basically need to utilise physics. If that is either by imitating your training data or actually understanding physics, is another matter. But as long as you have some kind of physics demonstration(video) in a tokenised format, then given a big enough dataset, your trained AI model will start having some emergent basic understanding of physics.

This is what OpenAI is doing because in order to generate a realistic video you need a simulation of the physical world. Well, would you guess, check this out..
Insane stuff
Please, Log in or Register to view URLs content!

Video generation models as world simulators​

We leverage a transformer architecture that operates on spacetime patches of video and image latent codes. Our largest model, Sora, is capable of generating a minute of high fidelity video. Our results suggest that scaling video generation models is a promising path towards building general purpose simulators of the physical world.

Turning visual data into patches​

We take inspiration from large language models which acquire generalist capabilities by training on internet-scale data.
Please, Log in or Register to view URLs content!
,
Please, Log in or Register to view URLs content!
The success of the LLM paradigm is enabled in part by the use of tokens that elegantly unify diverse modalities of text—code, math and various natural languages. In this work, we consider how generative models of visual data can inherit such benefits. Whereas LLMs have text tokens, Sora has visual patches. Patches have previously been shown to be an effective representation for models of visual data.
Please, Log in or Register to view URLs content!
,
Please, Log in or Register to view URLs content!
,
Please, Log in or Register to view URLs content!
,
Please, Log in or Register to view URLs content!
We find that patches are a highly-scalable and effective representation for training generative models on diverse types of videos and images.

Emerging simulation capabilities​

We find that video models exhibit a number of interesting emergent capabilities when trained at scale. These capabilities enable Sora to simulate some aspects of people, animals and environments from the physical world. These properties emerge without any explicit inductive biases for 3D, objects, etc.—they are purely phenomena of scale.
3D consistency.Sora can generate videos with dynamic camera motion. As the camera shifts and rotates, people and scene elements move consistently through three-dimensional space.
Interacting with the world. Sora can sometimes simulate actions that affect the state of the world in simple ways. For example, a painter can leave new strokes along a canvas that persist over time, or a man can eat a burger and leave bite marks.
Simulating digital worlds. Sora is also able to simulate artificial processes–one example is video games. Sora can simultaneously control the player in Minecraft with a basic policy while also rendering the world and its dynamics in high fidelity. These capabilities can be elicited zero-shot by prompting Sora with captions mentioning “Minecraft.”
These capabilities suggest that continued scaling of video models is a promising path towards the development of highly-capable simulators of the physical and digital world, and the objects, animals and people that live within them.
/End
 

canniBUS

Junior Member
Registered Member
The bar is definitely going up for China if it wants to stay in the game. Right now its generative Chatbots are not even available overseas and you need a Chinese phone number to even use Erniebot, for instance.

The claim that people overseas cannot access Chinese made chat bots is wrong. Right now anyone can download one of the open source Chinese made LLMs and run it on their own computer (or more likely on rented hardware).

Please, Log in or Register to view URLs content!
Please, Log in or Register to view URLs content!
Please, Log in or Register to view URLs content!

All of these models work as well as chat-gpt 3.5 and can approach chatgpt 4 in certain domains according to public benchmark measurements
Please, Log in or Register to view URLs content!
.
 

Overbom

Brigadier
Registered Member
The claim that people overseas cannot access Chinese made chat bots is wrong. Right now anyone can download one of the open source Chinese made LLMs and run it on their own computer (or more likely on rented hardware).

Please, Log in or Register to view URLs content!
Please, Log in or Register to view URLs content!
Please, Log in or Register to view URLs content!

All of these models work as well as chat-gpt 3.5 and can approach chatgpt 4 in certain domains according to public benchmark measurements
Please, Log in or Register to view URLs content!
.
Actually Qwen1.5-72b is imo the top open source AI model right now.

I am also looking forward to what Yi will release. Their Yi-34b model has been very impressive
 
Top