The New Sport of Misrepresenting AI Criticism
We recorded 8 claims from this post. 6 were checked against evidence. 2 can’t be checked. 3 more are held. Read the post at its source.
Claims from this post (8)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: UnresolvedSaid
“The argument I made there is that … systems like DALL-E don’t “understand what wheels are—or what they are for”, because (as I discovered on Sunday) they systematically fail to properly illustrate requests like these:”
What we tested: Systems like DALL-E do not understand what wheels are or what they are for, because they systematically fail to properly illustrate certain requests he tested.
Assessment: UnresolvedSaid
“The problem for the neural network community is not that I have changed my goal posts, it’s that my 2012 concerns still stand.”
What we tested: Marcus's 2012 concerns about deep learning still stand, and the problem for the neural network community is not that he has changed his goal posts.
Assessment: Can’t be checkedSaid
“It may be boring to hear about the same challenges over and over, but it’s hard to see how we will get to trustworthy general intelligence until these problems are resolved, in ways that are robust, rather than merely statistically contingent.”
What we tested: It is hard to see how we will get to trustworthy general intelligence until the challenges he has repeatedly raised are resolved in ways that are robust rather than merely statistically contingent.
Assessment: SupportedSaid
“Over and over I have stressed that you could have a good model of engineering without being a good model of humans, or vice versa.”
What we tested: Marcus has repeatedly stressed that a model could be a good model for engineering without being a good model of humans, or vice versa.
What happened: Faggella's interview report describes Marcus contrasting evolved human memory with more reliable computer memory. It records an earlier example of engineering desirability differing from human fidelity; the quotation preserves the source's apparent typo. (Emerj; Dan Faggella, )
Assessment: UnresolvedSaid
“Current neural nets lack the cognitive depth and robustness that would be necessary to support reliable L5 self-driving form vision alone.”
What we tested: Current neural networks lack the cognitive depth and robustness needed for reliable Level 5 self-driving from vision alone.
Assessment: SupportedSaid
“Over and over, the field of AI has discovered otherwise. Guy van den Broeck’s study is a perfect example of this.”
What we tested: The AI field has repeatedly found that doing well on a benchmark designed for X does not mean a system has actually solved X.
What happened: Zhang and colleagues, including van den Broeck, found near-perfect performance within a training distribution but substantial failures under different sampling of the same logical problem space. Their analysis attributes this to statistical features rather than acquisition of the intended reasoning procedure. (arXiv, )
Assessment: SupportedSaid
“But the smart money is still on being cautious about this, and in recognizing that what the Facebook researchers said in 2019 is still the case.”
What we tested: State-of-the-art models still learn to exploit spurious statistical patterns rather than learning meaning flexibly as humans do, so benchmark gains warrant caution.
What happened: Reducing misleading correlations improved independent-test performance for BERT and stronger RoBERTa and ALBERT models, supporting continued susceptibility in 2022. The same experiments found better generalization from larger models and debiasing, which limits any claim that no genuine progress occurred. (arXiv; ACL 2022 author manuscript, )
Assessment: Can’t be checkedSaid
“Those who wish that AI criticism would go away are welcome to their views, but misrepresenting critics and their views (and obscuring the degree to which even mainstream researchers agree) won’t magically solve the problems AI has yet to solve.”
What we tested: Misrepresenting AI critics and their views, and obscuring the degree to which even mainstream researchers agree, won't magically solve the problems AI has yet to solve.
Held from this post (3)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The two reviews read the claim differently
- The two reviews read the claim differently
- The two reviews reached different assessments. Our reading, not assessed: A recent article by Guy van den Broeck's lab shows that apparently successful performance on logic by neural networks is often driven by unstable statistical contingency.