discover the crazy project of Meta’s AI director

Meta’s prestigious AI director, Yann LeCun, wants to create an autonomous artificial intelligence. In a study, he unveils an innovative architecture allowing a machine to learn like a human being. Find out everything you need about this crazy project!

Over the past few years, artificial intelligence and machine learning have experienced unprecedented growth. At present, however, there is no no AI yet with human consciousness.

Recently, a former Google engineer was fired for claiming that a chatbot had become sentient. In reality we are still far from having reached this stage.

In comparison with humans or even with animals, current AI systems are not endowed with reason. However, this is an essential characteristic for the development of a truly autonomous machine.

Such an artificial intelligence would be able to learn directly from their observations of the real world, without the need for training sessions to accomplish a specific task.

However, for better or for worse, the first general-purpose autonomous AI could see the light of day soon. the Meta AI Director, Yann LeCuneminent figure in Machine Learning, has just unveiled his vision for a system capable of learning the way humans do.

Current artificial intelligence lacks common sense

In a study published by the Open site, Yann LeCun describes an autonomous AI capable of learning and experience the world similarly to a human.

His proposal is to train learning algorithms to learn more efficiently. Current AIs are not able to predict and plan real-world changeswhereas humans and animals can understand how the world works by observing it without even needing physical interaction.

For instance, teens can learn to drive a car in a few hours of lessons. They don’t need to experience a road accident on their own to achieve this.

On the contrary, current Machine Learning systems need to be trained on volumes of astronomical data to learn how to accomplish the smallest task.

As LeCun explained when presenting his work at UC Berkeley, “ a car with AI needs to fall off a cliff many times to understand that this is a bad idea. Then another few thousand times to figure out how not to fall off the cliff “.

The difference lies in “common sense”: a concept found in humans and the animal, but not in the current AI. In his study, LeCun describes common sense as a collection of patterns that can help a living being to tell the difference between what is probable, what is possible, and what is impossible.

It is this skill that allows a person to explore their environment, fill in missing information, and imagine new solutions for unknown problems. Although we take common sense as an innate skill, no scientist has been able to inject it into an AI or a Machine Learning algorithm.

Still during his speech, LeCun criticized learning techniques modern ones like learning by reinforcement or Reinforcement Learning. This training method consists of rewarding good behavior and punishing bad ones.

However, according to the expert, this approach will never match human reliability on real-world tasks. According to him, ” it is a concrete problem because we really want machines with common sense. We want self-driving cars, home robots, smart virtual assistants “.

An AI architecture similar to human memory

In order to initiate the advance of AI research for the next decade, LeCun proposes an architecture that minimizes the number of actions required by a system to successfully learn and complete an unassigned task.

the way that the different sections of the brain are responsible for the various functions of the body, LeCun suggests an artificial intelligence model composed of five separate and configurable modules.

One of the most complex parts of this architect would be there “ world model module » allowing to estimate the state of the world, and to predict imagined actions and sequences of other worlds in the manner of a simulator.

Using this unique world modeling engine, knowledge about how the world works could be shared very easily between different tasks. This architecture would therefore be similar to human memory.

However, significant efforts will be needed to create autonomous systems capable of coping with uncertain situations. This ability is nevertheless essential to evolve in a world so chaotic and unpredictable as ours

Yann LeCun: one of the founding fathers of modern AI

This is far from the first time that someone has mentioned the possibility of an autonomous AI, but Yann LeCun is not just anyone. Also professor at New York Universitythis specialist has spent most of his career developing learning systems on which many modern artificial intelligence applications are based.

In 2013 he founded Facebook group AI Research (FAIR) : Meta’s first breakthrough into the world of AI research. A few years later, he endorsed the role of chief AI scientist within the company.

Since then, Meta has had several significant successes in the field. In 2018, its researchers trained an AI to replicate eyeballs in hopes of simplifying digital photo editing.

In early 2022, Meta’s BlenderBot3 chatbot opened up the debate about AI ethics and biased data. And more recently, the firm unveiled its Make-a-Video tool capable of creating a video from text or an image.

By trying to give machines a better understanding of how our world works, Yann LeCun could open the door to a new generation artificial intelligence…

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *