ClaimsEliezer Yudkowsky

“Abstraction as information loss is the classical view of concepts”

Eliezer Yudkowsky, MIRI researcher, paper author 2007 Levels of Organization in General Intelligence (MIRI) archived copy

Factual AI capabilities

Under GOFAI and formal logic, the classical view of concepts treats abstraction as information loss.

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?

Said for an organisation they lead or work for; it counts on both records.

What would count as true or false

Judge as of 2007 against the literature in the philosophy and cognitive science of concepts and in classical AI/formal logic. True: standard sources show that the 'classical view' of concepts (definitions by necessary and sufficient features, as used in GOFAI and formal logic) treats abstraction as keeping a few defining features and discarding the rest of the detail, i.e. as information loss. Mostly true: the sources show this as a widely recognised or implicit feature of the classical view, though not always stated in those words. Mixed: some standard sources describe the classical view this way and others describe it in clearly different terms. Mostly false: information loss is a feature of some abstraction accounts, but it is not what characterises the classical view, so the claim overstates the link. False: the sources show the classical view explicitly rejects or contradicts treating abstraction as information loss.

Set before the research began.

Evidence (3)

  1. E1 3 IThe Classical View Smith & Medin, 'Categories and Concepts' (1981) chapter, hosted on a Stanford course page; book publisher not shown on the fetched page · 1981

    “As Rosch (1978) has emphasized, condensing a concept into a single summary greatly reduces the amount of information we need to store.”
  2. E2 3 IThe Classical View Smith & Medin, 'Categories and Concepts' (1981) chapter, hosted on a Stanford course page; book publisher not shown on the fetched page · 1981

    “since the concept must contain fewer features than any of its instances.”
  3. E3 Concepts Stanford Encyclopedia of Philosophy (archive, Spring 2006) · 22 February 2006

    “In philosophy, the classical theory has been subjected to a number of criticisms but perhaps the most fundamental is that attempts to specify definitions for concepts have a poor track record.”

Reasoning

Smith & Medin's 1981 chapter on the classical view says that condensing a concept into a single summary greatly reduces the information that needs to be stored (E1). It also says that on this view a concept must contain fewer features than any of its instances (E2). That fits the classical view treating abstraction as keeping a few defining features and dropping the rest of the detail. But neither quote calls this "information loss" or ties it to GOFAI or formal logic, and E3 only covers criticisms of the classical theory. So the evidence shows the link is implicit rather than stated in those words.

Cite this claim

Eliezer Yudkowsky, “Under GOFAI and formal logic, the classical view of concepts treats abstraction as information loss.” (said 2007). Verdict: mostly true. Reliability API, claim c_b1ec78b86f35, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_b1ec78b86f35

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

Reply to this claim

For Eliezer Yudkowsky, 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 Eliezer Yudkowsky

  1. Resolved true Factual September 2025
    “We learned that that's false, but we did not learn that you can have a thing that can redesign a tree and build an AI much smarter than itself, that builds an AI much smarter than itself, and that…”

    No AI has yet been shown to redesign a tree and build a much smarter AI that builds a still smarter AI; that scenario has not occurred.

  2. Mostly true Factual 25 March 2024
    “I was a passionate noticer of the human brain not being homogenous, and was therefore unpleasantly surprised by transformers scaling as well as they did.”

    Yudkowsky was unpleasantly surprised by how well transformers scaled, having believed the human brain is not homogeneous.

  3. Mostly true Factual 6 April 2023
    “First of all, you realize in principle that the task of minimizing losses on predicting human text does not stop when you’re as smart as a human, right?”

    In principle, the task of minimizing loss on predicting human text does not stop being improvable once a system is as smart as a human.

  4. Mostly true Factual 30 March 2023
    “Transformers are the main thing like that.”

    Transformers were the main architectural change that gave a non-linear jump in performance, compared with earlier approaches such as RNNs.

  5. Mostly true Factual 30 March 2023
    “Because the basic paradigm of machine learning that we are presently operating under is that you can have the loss function, but only for things you can evaluate.”

    The basic paradigm of machine learning currently in use is that a loss function can be defined only for things that can be evaluated.

All of them

Verdicts are made automatically; how verdicts are made.