Ruben C. Arslan

New work by @BjoernHommel and me. We fine-tuned a language model to predict correlations between survey items. In our pilot, the out-of-sample accuracy for item correlations was .71, .86 for reliabilities, and .89 for scale correlations. We're planning a preregistered follow-up.

Academic influencers hyping results because p < .05 😃 Actual academics:

Notes on an emerging legend - ChatGPT psychosis (a thread)

1. One clear takeaway is that expert forecasts and even user self-reports are not reliable indicators about AI capabilities. Actually Measuring Things is very valuable!

Indeed. This is not a good way of responding to criticism.

i think our arguments are more like precision instruments but ok fine i will accept "sledgehammer" (@DorsaAmir @anilkseth @guardian)

So funny. He wants you to believe in “brain drain” AND he has an app to sell you. x.com/rubenarslan/st…

I located the box & raccoon for this one years ago carcinisation.com/2020/01/27/ign…

in-depth investigations on the origin of Facts™ are crack to me

I'm reading Robert Sapolsky's Determined: Life Without Free Will as a mild form of self harm and here are two perfect pages - ego depletion, surgeon's birthdays, Francesca Gino/Dan Ariely study, implicit bias, hungry judges - all on the same page!!!

I helped @slatestarcodex with this post. At least with respect to some traits (e.g. education, bmi), we are in an epistemic crisis about their heritability. We have multiple strong methods - twin studies, RDR, sibling, GREML-WGS - that are not converging on a consensus reality.

i'm starting to think Scott has been quitely transitioning into becoming a comedian here. i've never felt so sinful finding myself laugh in tears 😂

Astral Codex Ten goes deep on the missing heritability question and recent findings on educational attainment astralcodexten.com/p/missing-heri…

These results are inconsistent with similar previous analyses such as this from our education GWAS paper. For IQ, the PGS effect shrunk to 84% as large (S.E. 3.3%) within-family. We used a much larger sample than this study, meta-analysing across UK Biobank, Swedish Twin…

Has nominative determinism gone too far (h/t @bschne)

I think there are some broader implications for evolutionary psychology here. As I try and emphasize all the time now, EP is an important and necessary field that helps shed light on human nature. There is lots of good work in it, plenty of people who have promoted the ‘gender…

Interesting new paper showing what I’ve been complaining about for years: the ‘gender-equality paradox’ is confounded by spatial and cultural autocorrelation: “gender differences covary more strongly with cultural regions and data quality than gender equality”.

study 1 of the infamous "sign at the top" study claims to test how much signing at the top or bottom of a form affects cheating at a math task...but apparently the subjects didn't even see the form they signed until AFTER the cheating task??

the most infuriating study is trolling me personally from 2014