Post

They keep trying to make "correlation is cognition" happen. Through asserting: 1️⃣ models with only correlationist powers (data and correlations over that data) work well & 2️⃣ our own cognitive powers are no different to these machines, which can only rehash the past. @andreaeyleen.eurosky.social 1/

Mark Dingemanse@markdingemanse.net

(Guest & Martin, 2023:24) link.springer.com/article/10.1...

68 reposts225 likes

They do this in a variety of ways, here's one tactic that directly hits the public through journalists and technology company logics. 2/

This hangs with the other statements in fig 1: 1️⃣ ANNs (artificial neural networks) are inspired by human organism 2️⃣ ANNs are a black box we don't understand 3️⃣ We understand neither ANNs nor cognition therefore they are similar All 3 are misplaced! We unpack this... doi.org/10.5281/zeno... 3/

Another is within cognitive scientific reasoning itself. Also with @andreaeyleen.eurosky.social... where we describe how this manifests. 3/

Another, related aspect, that I wanted to highlight in this work is the difference in stances — some much more damaging than others — that appears to have occurred around 2010. These are summarised in Table 1, here: doi.org/10.1037/rev0...; e.g. goals seem to shift from theory to prediction. 3/n

They also do this by compressing distortion and damaging the pre-existing meaning of AI, which while problematic is now turbocharged with harms. To fix this we need to do something to directoy address the malformed logics people adopt through thinking AI is a sensible technology. It's not! 4/

I'm really glad more and more people are noticing correlationism, part of the conspiratorial logic of AI... Guest, O. & Martin, A. E. (2025). A Metatheory of Classical and Modern Connectionism. Psychological Review. doi.org/10.1037/rev0... PDF: repository.ubn.ru.nl/bitstream/ha...

For the meaning of AI: I actually cringe when people think there's something deep to argument from ignorance — that because it's a vague term that something follows... Nothing follows except if we want to refine the term or its referents. 5/

I split AI into 3 non-mutually exclusive types (see Table 1 above): displacement (harmful), enhancement (beneficial), and/or replacement (neutral) of human cognitive labour. More later possibly, but see Tables 2 to 4 (attached or here: arxiv.org/pdf/2507.19960) for the worked through examples. 2/n

I just want to assert that there are plausible positions between "it's just fancy autocomplete" and "it has feelings just like me". That's all really. Thanks for your thoughts everyone.

While I agree that AI is over-hyped, the "it's only correlation" argument ignores higher-level, emergent effects. It's like saying Simone Biles's performance is only Newtonian mechanics.

Seems like similar bad reasoning as information-processing models: If you have the same inputs and get the same output, then the process in between must be identical. Was wrong then and is wrong now.