Close Menu
Versa AI hub
  • AI Ethics
  • AI Legislation
  • Business
  • Cybersecurity
  • Media and Entertainment
  • Content Creation
  • Art Generation
  • Research
  • Tools

Subscribe to Updates

Subscribe to our newsletter and stay updated with the latest news and exclusive offers.

What's Hot

Creating innovative content at your fingertips

July 4, 2025

The UK and Singapore form an alliance to guide AI into finance

July 4, 2025

StarCoder2 and Stack V2

July 4, 2025
Facebook X (Twitter) Instagram
Versa AI hubVersa AI hub
Friday, July 4
Facebook X (Twitter) Instagram
Login
  • AI Ethics
  • AI Legislation
  • Business
  • Cybersecurity
  • Media and Entertainment
  • Content Creation
  • Art Generation
  • Research
  • Tools
Versa AI hub
Home»Research»Open source, Deepseek, Meta, and Model Research
Research

Open source, Deepseek, Meta, and Model Research

By January 29, 2025Updated:February 13, 2025No Comments6 Mins Read
Share Facebook Twitter Pinterest LinkedIn Tumblr Reddit Telegram Email
Share
Facebook Twitter LinkedIn Pinterest Email

Neuron system with a glowing connection on a black background

Getty

Asking 10 people how the large -scale movement of the language model works, you can get 10 answers. Or you may actually get only two or three answers. The remaining respondents lament the lack of general understanding of these extensive “courage”, saying, “I don’t know, don’t worry.” system.

It is rarely encountered such a ubiquitous technique, recognizing that the majority of people really do not know how it works. However, for fairness, this stage was already set for this kind of knowledge gap in the world of technology. Many people explained what the cloud was 10 or 15 years ago. Regarding cipher, most people still hate to explain the blockchain in very detail.

However, AI is different. Furdies are different -the impact of our society and private life is different. Therefore, it will help you know how AI agents, LLMS, and neural netets make decisions and process them around them.

I go to some basic elements of hearing that people are talking about using some specific quotes from the famous computer scientists Yann Lecun. 。 Lecun had some very important things about the actual state of AI development and what it means for our world.

Computer collection

This is the first main idea that people close to the industry have presented to the audience for the past few years.

In a sense, it is an exemption on the rapid appearance of AI on behalf of a kind of human. This is a general qualifier that AI has no obstacles to acquire more human beings over time.

If you look at the results of a large new model like Openai O1 and the latest models of DeepSeek, you may think that AI can jump on human intelligence rods simply by adding power and tokens.

But, according to some of the best theorists, that would be wrong. Because true intelligence is a collection of systems that cooperate to produce very accurate and robust results.

In Marvin Minsky’s “Society of the Heart”, famous theorists have a wide variety of statements about artificial intelligence in contrast to the human mind. Human hearts are not one computer, but many small interconnected agents that make up a “society of the heart”, and all individual components cooperate with greater results.

Think about it: Social people are working together for a consistent purpose. It is a model given by Minsky -a very powerful supercomputer not only becomes a god in its intelligence, but also a “village” with an interconnected or module like living cells.

Yann Lecun reflects this when commenting on the development of LLMS.

“We are not a place where we cannot reproduce the kind of intelligence that can be observed by animals, not humans,” he says. “Intelligence is not just linear. Here, over the barrier, there are human intelligence and super intelligence. It is not at all. It is a collection of skills and gaining new skills very quickly. And the ability to solve the problem without actually learning anything.

When you start thinking about such AI, you change the way to frame it, which creates a big difference.

Memory and context

Lecun also suggests that “permanent memory” is needed for these systems to be powerful. And that’s what we have talked to us over and over again when an expert is working on a new AI model.

This can be spoken in a context window used to perceive what the system is working on, for example, from the viewpoint of memory or from context.

To do this, let’s talk about two types of memory, parametric memory and working memory.

These two types of memory can be mapped to two different types of human memories that make up short -term memory and long -term memory (short -term memory and long -term memory). Parametric memory is a long -term memory, that is, historic context and knowledge that has long -term thinking of machines. Working memory is a short -term memory. This basically corresponds to the clues for context windows and contexts where the machine learns in real time. (See the details of this resource of the Data Scientist Association)

Both of them have their own importance in the AI ​​system, as they do in the human brain. Neurologists talk about the differences in long -term and short -term human memory functions. The theorist should talk about the difference between AI’s parametric memory and work memory.

New framework

As Minsky and Lecun pointed out, what is taking home if the human mind is more elaborate than a single supercomputer?

“I need a completely new architecture,” Lecun says. “It doesn’t happen with LLMS. Nothing happens so far. We need to have common sense … If we give the LLM a standard puzzle, the answer will flow back.”

One of his proposals is the emergence of what is called the “World Model,” and the AI ​​begins to build the necessary context and become more recognized as humans.

“What happens if you have some idea about the state of the world and imagine the actions you might take?” “Can you predict the next state of the world from this action? … If you have such a system, you may be able to predict what action sequence will create. Action that satisfies the task. They are this purpose. That’s what we’re working on … and maybe some specific results in a wider world within a few years. The year is almost more difficult than we think.

Open source model advantage

This does not end without mentioning Lecun’s current statement about the open source model.

The business world (and investment world) is suitable for new announcements of Deep Seek, a Prime Open Source AI model that postpon plans by US companies like a meta. However, instead of regarding it as “the United States to China,” Lecun urges it to consider it as its own open source. And it also helps our expectations for what will happen after this year.

Apply these frameworks to AI research when moving forward.

author avatar
See Full Bio
Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
Previous ArticleAnalog in memory computing may strengthen tomorrow’s AI model.
Next Article Deepseek’s hidden warning to AI safety

Related Posts

Research

In the midst of intense AI talent races, Meta’s active recruitment target open-rai researcher

June 30, 2025
Research

Lossless compression tailored to AI

June 30, 2025
Research

High-tech research jobs in the US will rise by 26% by the next decade. Median future salary for AI, ML and others is $140,000

June 30, 2025
Add A Comment
Leave A Reply Cancel Reply

Top Posts

New Star: Discover why 보니 is the future of AI art

February 26, 20252 Views

Impact International | EU AI ACT Enforcement: Business Transparency and Human Rights Impact in 2025

June 2, 20251 Views

Presight plans to expand its AI business internationally

April 14, 20251 Views
Stay In Touch
  • YouTube
  • TikTok
  • Twitter
  • Instagram
  • Threads
Latest Reviews

Subscribe to Updates

Subscribe to our newsletter and stay updated with the latest news and exclusive offers.

Most Popular

New Star: Discover why 보니 is the future of AI art

February 26, 20252 Views

Impact International | EU AI ACT Enforcement: Business Transparency and Human Rights Impact in 2025

June 2, 20251 Views

Presight plans to expand its AI business internationally

April 14, 20251 Views
Don't Miss

Creating innovative content at your fingertips

July 4, 2025

The UK and Singapore form an alliance to guide AI into finance

July 4, 2025

StarCoder2 and Stack V2

July 4, 2025
Service Area
X (Twitter) Instagram YouTube TikTok Threads RSS
  • About Us
  • Contact Us
  • Privacy Policy
  • Terms and Conditions
  • Disclaimer
© 2025 Versa AI Hub. All Rights Reserved.

Type above and press Enter to search. Press Esc to cancel.

Sign In or Register

Welcome Back!

Login to your account below.

Lost password?