Adam Kucharski

Is it worse to believe too much or doubt too much?

Launching on YouTube tomorrow! www.youtube.com/@adamjkuchar...

Science and health are screwed. Or are they? Latest post: kucharski.substack.com/p/science-is...

I used to think that people use the “weasel word causal inference strategy” (“our analyses show that X is an important predictor of…” followed by conclusions that are clearly causal) for the sake of plausible deniability. But I’m revising that now, because often deniability *isn’t* plausible.>

A bit of a Christmas mystery What looks strange about £21.5m?

What is strange about the value £21.5 million? My piece on a runaway algorithm, a Christmas mystery and a great bit of reverse engineering insight: kucharski.substack.com/p/a-bit-of-a...

“We reveal a troubling trade- off: When AI systems are optimized for persuasion, they may increasingly deploy misleading or false information.”

www.science.org/doi/epdf/10....

🚨 New in Nature+Science!🚨
AI chatbots can shift voter attitudes on candidates & policies, often by 10+pp
🔹Exps in US Canada Poland & UK
🔹More “facts”→more persuasion (not psych tricks)
🔹Increasing persuasiveness reduces "fact" accuracy
🔹Right-leaning bots=more inaccurate

YouTube video by Adam Kucharski Launching soon!

Arriving Monday! Subscribe now to keep updated: www.youtube.com/watch?v=LZbr...

Latest post, testing out Nano Banana on my son’s adventures… kucharski.substack.com/p/nano-banan...

There's probably a more elegant method I'm missing though!

Without causing a full blown geopolitical crisis, my wife is half-dutch so it's neither...

My son is having a public health strategy for his packed lunch today.

If you're an expert in your field but still occasionally feel a tinge of imposter syndrome, just have a look at some of the current ACIP lineup.

1/2

Share your Christmassy Science!

To mark Proof being an FT Book of the Year, I’m giving away 2 copies in run up to Xmas – prize will go to the quirkiest, funniest, or most interesting Christmas-related science.


Comment below – most likes by 8th Dec wins (Europe/US only).

To kick things off, here's Galton's 1906 Nature letter:

Just did it geometrically – write a quick R script to start at the peak point and work out where it hits the GAM confidence interval on either side.

Had a quick go at fitting a GAM and showing the 95% interval for each 'peak':

Noisy data + single smoothing curve = overly precise conclusions about where the peaks land.

www.cam.ac.uk/stories/five...

And, of course, we also need confidence that appropriate data will be promptly collected and made available for estimation and/or estimates available for reuse in downstream analysis.

Don't think anyone wants to be digging numbers of a PPT or PDF in the middle of a pandemic.

Good blog which raises some of the questions that the inquiry seems sadly to be missing.

The key point for me is towards the end: if we conclude from COVID-19 that pandemic preparedness is about epidemiological models, PPE, or healthcare capacity, we weren't paying attention.

The UK isn’t prepared for the next pandemic – and nor are many other countries.
kucharski.substack.com/p/we-arent-p...

It's remarkable how many of the negative online reviews for Proof are because I describe the Jan 2021 riot at the Capitol building as, well, a riot.

You know, that riot we all watched happening in real-time.

There are some serious and strong attempts at revisionism at play these days.

YouTube video by Adam Kucharski Launching soon!

How can we make sense of a noisy, uncertain world?

I’ve spent lots of time writing online about this topic, and speaking in person. So I thought it was time to combine the two…

Excited to launch soon, thanks to support of TED Fellows.

Subscribe now to stay updated: www.youtube.com/@adamjkuchar...

Fair point – more downloaded than read.

Can everyone just stop posting muddled Nano Banana infographics of papers they've read?

Share what *you* think about it!

YouTube video by Adam Kucharski Launching soon!

How can we make sense of a noisy, uncertain world?

I’ve spent lots of time writing online about this topic, and speaking in person. So I thought it was time to combine the two…

Excited to launch soon, thanks to support of TED Fellows.

Subscribe now to stay updated: www.youtube.com/@adamjkuchar...

It's lucky the UK doesn't have any key industries or services reliant on this kind of immigration.

*checks notes*

Oh...

To mark Proof being an FT Book of the Year, I’m giving away 2 copies in run up to Xmas – prize will go to the quirkiest, funniest, or most interesting Christmas-related science.


Comment below – most likes by 8th Dec wins (Europe/US only).

To kick things off, here's Galton's 1906 Nature letter:

Reminds me of that quote that your inbox is someone else's to-do list.

Making emails much easier to write and send with AI will make us all more productive, will it?

My inbox says otherwise...