ClaimsYann LeCun

“shallow learning systems have one or two layers, while deep learning systems typically have five to 20 layers.”

Yann LeCun, interviewee 18 February 2015 Facebook AI Director Yann LeCun on His Quest to Unleash Deep Learning (IEEE Spectrum) archived copy

Factual AI capabilities

Shallow learning systems have one or two layers, while deep learning systems typically have five to 20 layers.

Mostly true True in substance, wrong in a detail.

Factual: Statements about the present or the past, checked against the record: did they hold up?

The second judge would not resolve this claim; the verdict shown is the first judge's.

What would count as true or false

Judge as of 2015-02-18 against how machine learning literature of that time described layer depth. True: literature of about 2015 generally treats shallow models as having one or two layers (e.g. linear models, kernel SVMs, single-hidden-layer networks), and typical deep networks then in use (e.g. AlexNet, VGG, GoogLeNet and contemporary speech models) mostly fell within about 5–20 layers. Mostly true: both characterizations broadly hold, but the typical deep range in common use was somewhat different (e.g. commonly up to 22–30 layers) or the shallow boundary was drawn slightly differently. Mixed: one of the two characterizations holds and the other clearly does not. Mostly false: both are significant distortions of usage at that time. False: typical deep systems in 2015 were plainly outside this range (e.g. mostly hundreds of layers, or mostly 2–3 layers).

Set before the research began.

Evidence (4)

  1. E1 Going deeper with convolutions arXiv (Szegedy et al., Google) · 15 September 2014

    “The network is 22 layers deep when counting only layers with parameters (or 27 layers if we also count pooling).”
  2. E2 Going deeper with convolutions arXiv (Szegedy et al., Google) · 15 September 2014

    “However, contrary to the fixed 2-layer deep model of [15], all filters in the Inception model are learned.”
  3. E3 Highway network - Wikipedia Wikipedia · 28 September 2017

    “In 2014, the state of the art was training deep neural networks with 20 to 30 layers.”
  4. E4 Deep learning — Marovi Marovi · 31 October 2026

    “In practice the word marks networks with more than one or two learned layers”

Reasoning

E4 says "deep" refers to networks with more than one or two learned layers, which fits the claim's line between shallow and deep. On the upper end, E1 says the 2014 network in 'Going deeper with convolutions' had 22 layers counting only layers with parameters (27 including pooling), and E3 says the 2014 state of the art was 20 to 30 layers. So the cutting-edge deep models sat at or slightly above the claimed 5–20 range. That matches the criteria's "mostly true" case, where typical deep networks commonly reached 22–30 layers.

Cite this claim

Yann LeCun, “Shallow learning systems have one or two layers, while deep learning systems typically have five to 20 layers.” (said 18 February 2015). Verdict: mostly true. Reliability API, claim c_3717e530daad, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_3717e530daad

This claim is in dataset release v1.1.2, with the same verdict.

Reply to this claim

For Yann LeCun, or someone speaking for them: a correction or counter-evidence. Replies are read before anything is shown; an approved reply shows your name, role, date, statement and link. Your email address is never shown and is deleted 30 days after the decision (privacy).

More factual claims by Yann LeCun

  1. Mostly true Factual 18 March 2024
    “For starters, transformers are equivariant to permutations.”

    Transformers are equivariant to permutations, as ConvNets are equivariant to translations.

  2. Mostly true Factual 14 March 2024
    “If it were the case, we would have AI systems that could teach themselves to drive a car in 20 hours of practice, like any 17 year-old.”

    There are currently no AI systems that can teach themselves to drive a car in about 20 hours of practice, as a 17-year-old can.

  3. Mostly true Factual 8 March 2024
    “The architecture is good, the architecture of the encoder is good, but the fact that you train the system to reconstruct images does not lead it to produce to long, good generic features of images.”

    Training an encoder to reconstruct images does not lead it to produce good generic image features, though the encoder architecture is good.

  4. Mostly true Factual 8 March 2024
    “LLMs can do none of those or they can only do them in a very primitive way and they don’t really understand the physical world.”

    Current LLMs can do none of understanding the physical world, persistent memory, reasoning and planning, or only in a very primitive way.

  5. Mostly true Factual 8 March 2024
    “No, there’s one thing that autoregressive LLMs or that LLMs in general, not just the autoregressive one, but including the bird style bidirectional ones, are exploiting and its self supervised…”

    LLMs, including BERT-style bidirectional ones, exploit self-supervised learning, and LeCun has been a very strong advocate of self-supervised learning for many years.

All of them

Verdicts are made automatically; how verdicts are made.