Chris Chapman

Same! Not sure I can achieve 100% but have been trying for a few weeks.

FWIW, so far today I have given a hard pass to commenting or sharing 2 different Substack posts 😞

This is again a reminder that college isn’t supposed to just teach you a job but is supposed to teach you how to think, because that job you’re training for might not exist when you graduate

Photo of me holding a printed copy of my book

Draft cover art

Two -page spread showing graphs text and code

Two -page spread showing graphs text and code

#rstats So chuffed! I printed the first PDF copy of my book to see what it might look like in print! It will be printed in full color!

Visualizing Multivariate Data and Models in R

On the whole, looks good, but lots of tweaking to do.
It weighs in at ~440 pgs., so perhaps some cutting needed.

Social media researchers: you can't test the effects of UX tweaks on user behavior by populating a synthetic platform with virtual agents instantiated by LLMs *that have been trained on social media platforms without these tweaks*

"PhD-level experts in your back pocket" is a completely nonsensical description of AI but a pretty good description of social media if you follow the right people

Very real question I have for folks bringing AI into their classrooms proactively (from a place of 100% curiosity!):

How are you navigating students who have explicitly named their discomfort or unwillingness ethically with the technology?

Astonishing and dismaying to see how many otherwise sensible people are willing essentially to say "okay, this tool doesn't correctly answer a question I know the answer to, but look how amazing it is at answering these questions I don't know the answer to" without seeing the problem here.

I call this the reverse Turing Test. If an LLM is producing something passable, it doesn't mean the tool is intelligent, it means humans were being asked to write something meaningless

Surveys and the "True Score" Mistake Increasingly, I think there is a deep and mostly unstated assumption about survey research that drives profound misunderstanding: the belief in a "true score". Researchers and their audiences expect that there is an underlying real value that we can ...

Among many things, reasons to take surveys (e.g. the message they want to be heard) could be changing.

The idea of "latent scores" for surveys is somewhere between unlikely and outright wrong, as I discuss here: quantuxblog.com/surveys-and-...

"You can see the computer age everywhere but in the productivity statistics". --Solow, 1987

Again, you cannot unplug 1/4-1/5 of an R1 university's budget & expect the same quantity or quality of "things" on campus. The math simply doesn't work. And pretending it does & we can uphold our "core values," & not talking loudly & publicly about how it doesn't, is not a strategy. It's madness.

Galaxy brain at work. When the goal is to "mimic natural immunity" then every placebo is already effective!

George Carlin maniac vs idiot, AI version!

One of the most common type of AI post in UX is a booster pretending to be a skeptic: "AI promised it would be simple but to use it well you actually need very good design taste, like I have"

Managers don't give a shit about your "good taste" because *they don't know their AI outputs are bad.*

One reply I've heard too often is, "But can you PROVE this isn't good enough??"

You only have to use it very, very carefully, right?

Not like all those other dummies (namely everyone except oh-so-careful me).

Don't Be Fooled By Trump's Plan To 'Upskill' Workers To Prepare For AI | TechPolicy.Press The future of work does not require an unquestioning acceptance and embrace of AI, write DAIR's Alex Hanna, Tina Park, and Sophie Song.

Tina Park, Sophie Song, and I wrote about Trump's AI Action Plan, the narrative of "upskilling" on both the right and left, and how the myth of AI automation is forcing actual workers to do more with less.

www.techpolicy.press/dont-be-fool...

An algorithm is just an algorithm ... but I worry for both the users who read such results -- and those who wrote whatever went into the training corpus

How the Quant UX book was written This post shares high-level technical details of how Kerry Rodden and I wrote the Quant UX book. I start by describing the overall process and then I review how it all comes together in RStudio. (Another time I'll discuss more about the process of de...

One of my books was written in R markdown. Literally sent the PDF directly to the publisher 💥

Not quite camera ready, but very close. They just tuned it up a bit for their house style.

Details at quantuxblog.com/how-the-quan...

Illinois is the first state to ban AI therapists Illinois Governor JB Pritzker has signed a bill banning AI therapists into law, the first in the nation.

Illinois has banned AI therapy, making it the first state to regulate the use of AI in mental health services.

www.engadget.com/ai/illinois-...

there is almost nothing I can think of that’s more enriching to kids’ lives than creating their own imaginary worlds on paper or canvas or clay or screen, and anything you do to shortcut that process to prevent them from learning how to do that kind of creative thinking hurts them forever. FOREVER.

Ghost 6.0 Networked publishing, native analytics, and $100M earned by indie publishers

Gotta say, the new Ghost 6.0 update is exciting (Substack alternative for newsletters).

A realtime web traffic dashboard, integration with the Fediverse, etc.

If you are thinking of leaving Substack, there's never been a better time. It's less expensive!

ghost.org/changelog/6/

amongst current misunderstanding and simplistic understanding of AI, the popular narrative of "bias in, bias out" continues to frustrate me the most. this thread covers some of the reason

US election data nerds: Associated Press is hiring 3 election data analysts and an election data scientist in DC. I don't know more than seeing the listings.
careers.ap.org/job/Washingt...

careers.ap.org/job/Election...

#ddj

The raw milk people are hilarious. All you have to do is meet one (1) cow and smell the situation in which they are producing that rawness and you will pasteurize everything you own for a week

Congratulations, thank you, and subscribed!

Wanna see a crazy chart? Of course you do.

Spending on constructing datacenters and manufacturing facilities for computers & electronic products in the US now exceeds the entire retail and wholesale industry and almost as much as the rest of the office and manufacturing sectors combined.

🧵, 🥺

Increasingly I think a major function of “masculinity” IS its crisis — a vehicle for the dominant & dominating group to justify focusing their social attentions primarily on themselves

Photo from The Guardian previewing a story. It shows a photo of flooding in India, with a headline to the main story that says "A volcanic eruption, floods in India and Hiroshima"

In The Guardian today: a reminder of why Oxford commas are the right choice.

Namely, it is not flooding in Hiroshima!

www.theguardian.com/news/gallery...