Meta’s Yann LeCun predicts ‘new AI architecture paradigm’ and ‘decade of robotics’ within 5 years

Meta’s Chief Artificial Intelligence Scientist, Yann LeCunsays that in the next three to five years, a “New paradigm of artificial intelligence architecture” will emerge, far beyond the capabilities of existing artificial intelligence systems.

LeCun also predicted that the coming years could be the “decade of robotics,” where advances in artificial intelligence and robotics combine to open up a new class of smart applications.

speech at the sessionDebate Technology” In Davos on Thursday, LeCun said that the “flavor of AI” we have now, namely generative artificial intelligence and large language models (LLM), is not really that much. Useful, no doubt, but it falls short in many ways.

“I think the current (LLM) paradigm has a fairly short shelf life, probably three to five years,” LeCun said. “I don’t think anyone in their right mind will be using them, at least as a central component of an AI system, in five years. I think (….) we will see the emergence of a new paradigm for AI architectures that may not have the limitations of existing AI systems.”

LeCun says these “constraints” prevent truly intelligent behavior in machines. There are four main reasons for this: lack of understanding of the physical world; lack of persistent memory; lack of justification; and lack of comprehensive planning capabilities.

“LLMs don’t really do any of that,” LeCun said. “So there will be another revolution in artificial intelligence over the next few years. “Maybe we should rename it, because it probably won’t be generative in the sense we understand it today.”

“World Models”

It echoes LeCun’s sentiments married in the past. Underlying it are things to be known as.world models” that promises to help machines understand real-world dynamics. This includes having memory, common sense, intuition, reasoning—characteristics very different from existing systems, mostly related to pattern recognition.

Earlier, LeCun said it could still happen for about 10 yearsbut today’s estimate puts things even closer on the horizon. Although it is not clear how far it will reach during this period.

“LLMs are good at manipulating language, but not at thinking,” LeCun said. “So that’s what we’re working on — systems building mental models of the world. If the plan we’re working on succeeds on the schedule we hope for, we’ll have systems that are a completely different paradigm within three to five years. They may have some level of common sense. “They can learn how the world works by observing and perhaps interacting with it.”

“Decade of Robotics”

as impressive is generative artificial intelligenceis able pass the bar exam or drilling new drugsLeCun believes robotics could be a central component of the next wave of AI applications in such real-world scenarios.

The meta itself does some research work in the field of roboticsalso an AI lover, ChatGPT creator OpenAI. Earlier this month new job postings have appeared details the new OpenAI robotics team, which focuses on “general-purpose,” “adaptive,” and “versatile” robots with human-like intelligence in real-world settings.

“We don’t have robots that can do what a cat can do – understanding the physical world of a cat is far beyond anything we can do with artificial intelligence,” he said. “Maybe the next decade will be the decade of robotics, maybe we’ll have AI systems that are smart enough to understand how the real world works.”