Skip to content
TheChatArchiveThe Chat Archive
Perplexitysonar-reasoning-pro-high· PerplexityHugging Face LMArena Search Arena 24K

explain why yann le cunn says llms are on their way out. give specific quotes

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Perplexitysonar-reasoning-pro-high’s answer.
  1. User

    explain why yann le cunn says llms are on their way out. give specific quotes

  2. Perplexity Assistant

    Yann LeCun, Meta's Chief AI Scientist and a Turing Award laureate, has made several strong statements about the limited future of Large Language Models (LLMs). His view is that current LLMs represent a technological paradigm that will soon be obsolete.

    Prediction of LLMs' Short Lifespan

    LeCun has made explicit timeframe predictions about how long the current LLM paradigm will last:

    "I think the shelf life of the current [LLM] paradigm is fairly short, probably three to five years. I think within five years, nobody in their right mind would use them anymore, at least not as the central component of an AI system."[1]

    He elaborated further in a recent interview with Newsweek, stating: "If we can achieve our goals within three to five years, we will establish a vastly improved framework for systems capable of reasoning and planning."[2]

    Fundamental Limitations of LLMs

    Lack of Essential Intelligence Characteristics

    LeCun identifies four key components necessary for true intelligence that LLMs fundamentally lack:

    "There is a number of characteristics of intelligent behavior. For example, the capacity to understand the world, understand the physical world, the ability to remember and retrieve things, persistent memory, the ability to reason and the ability to plan. Those are four essential characteristic of intelligent systems or entities, humans, animals. LLMs can do none of those, or they can only do them in a very primitive way."[5][6]

    Language vs. World Understanding

    A central critique is that LLMs focus on language patterns rather than understanding the world those patterns describe:

    "The goal is for AI systems to focus on the world being talked about, not the words themselves — but LLMs don't grasp the distinction."[3]

    He explains that this creates a shallow understanding:

    "LLMs' understanding of language, while impressive, is shallow."[3]

    Limited Reasoning Capabilities

    In describing how LLMs process information, LeCun states:

    "An LLM generates one token at a time, processing a fixed amount of computation to produce each token. This is clearly a reactive process—there's no reasoning involved."[2]

    He compares this to Daniel Kahneman's framework:

    "LeCun posits that today's AI systems are predominantly System 1... without engaging in broader reasoning."[4]

    Inadequacy of Text-Only Training

    LeCun argues that training solely on text data has fundamental limits:

    "A large LLM today is trained on approximately 10^14 bytes of training data, which take any us 170,000 to read through. This clearly indicates that we will never achieve human-level intelligence by solely training on text. It's simply not going to happen."[2]

    He contrasts this with the richness of human learning:

    "The optic nerve bandwidth is around one megabyte per second. Multiply that by 16000 hours and you arrive at about 10^14 bytes in four years instead of 400,000."[2]

    Future Role of LLMs

    While LeCun predicts LLMs will fade as the central AI approach, he still sees a limited role for them:

    "There's a minor function for LLMs, primarily in translating abstract thoughts into language. In the human brain, this task is managed by Broca's area, located here. This area only emerged in the last couple of hundred thousand years. If you lose the function of Broca's area, you can still think; you just can't articulate your thoughts."[2]

    The Coming AI Revolution

    LeCun predicts a paradigm shift:

    "So there's going to be another revolution of AI over the next few years. We may have to change the name of it, because it's probably not going to be generative in the sense that we understand it today."[1]

    Instead, he envisions systems with "authentic representations of reality, allowing users to set objectives that the systems can pursue effectively."[2]

    This evolution is expected to create AI with more human-like abilities to understand physical reality, reason, maintain persistent memory, and plan effectively – capabilities that current LLMs fundamentally lack.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.