@cchapman
UX researcher, psychologist. Author "Quantitative User Experience Research" (w/Rodden), "R | Python for Marketing Research and Analytics" (w/Feit & Schwarz). Previously 24 yrs @ Google, Amazon, Microsoft. Personal account. Blog at https://quantuxblog.com
A common question I get about LLMs is whether I use them to write code.
My answer is "no" ... and in this outstanding discussion, @jacobharr.is explains why not, FAR better than I could.
I could add reasons for myself (e.g., aesthetics and plagiarism) but overall this is the best answer I've seen.
GOOD NEWS! Researchers at Stanford University have developed a UNIVERSAL vaccine known as GLA-3M-052-LS+OVA, that protects against a wide range of respiratory viruses, bacteria AND even allergens. The vaccine is delivered intranasally AND provides broad protection in the lungs for several MONTHS.
Yikes 😳 So many red flags in a single diagram:
Ground Truth
Partner Provided Ground Truth
Standardized Cognitive Tests
Quality Assessors
"Model" is on par with "Learner"
Learner Response Signals
Learning Outcomes
... and that's before digging into the problematic path connections and implications
Outstanding 10 month old thread on LLMs (two levels down from this post).
In addition to being an excellent summary, it covers the typical questions, "but what about ... [coding]?" etc.
Not very surprised here 🤷 As a general matter, I've argued that the entire concept of synthetic polling data has multiple fundamental flaws.
Video webinar (free) and PDF slides both available: sawtoothsoftware.com/resources/ev...
Vibe coding broke people's brains because they had a bad understanding of the software design process.
The pop culture model goes something like this: start out with a sketch and then render that same idea in progressively finer detail.
So when tools could "skip" to high detail, people went WOW.
Mostly agree, but a couple of exceptions.
Green Apple books at SFO and Powell's at PDX (Portland) are quite nice (considering 'airport').
Sadly, Elliott Bay Books at SeaTac is no more.
Postdoc job! I'm looking to recruit a postdoc to work on a project about research evaluation & metadata. More info here: www.yorku.ca/research/wp-...
Salary: CDN$70,000 (+ benefits)
Location: Toronto, ON (can be remote in ON)
Length: 18 months
Deadline: 31 March 2026
Start date: 1 July 2026
Not only are book clubs being ruined (by plagiarism machines!) ..
.. as a nonprofit organizer, I'm inundated with AI emails about accounting services, web design, call center "help", VC funding, on and on.
Just today a crucial service is raising prices because of "new AI features". Maddening.
Coworker and I were discussing colorblind palettes and grayscale, and I made a blog post with some #rstats resources: cmep-ms.github.io/blog/posts/2...
Most of the footage should never leave the user’s device. But of course the worst company doesn’t do privacy.
Meta Workers Say They’re Seeing Disturbing Things Through Users’ Smart Glasses
futurism.com/artificial-i...
🦊
We cannot meaningfully separate everyday usage of “AI” from the machine of death and war. If you’re upset by the military-industrial complex? These tools feed it; they *are* it.
ethanmarcotte.com/wrote/propel...
A picture of PEW poll on global attitude survey, saying the % who rate the morality and ethics of people in their country as good vs bad, where the US has the worst rankings and Canada the best
Americans: we live in a fallen state—embroiled by sin, cheating, lying, and evil. You cannot trust anyone, not even those who claim to know you best
Canadians: I love my neighbors and my friends!
bsky.app/profile/stat...
Same here, every page I've tried says something went wrong or (common, mass) product is unavailable.
Also reports say Prime Video is down.
New article by Britt Paris, Serita Sargent and me in Big Data & Society
We discuss collective bargaining around AI and how organized labor can meaningfully push back on—and even refuse—the pervasive narrative of AI inevitability
journals.sagepub.com/doi/10.1177/...
This is a good principle for design in general:
When users tell you there's a problem, they are usually right.
When users tell you how to fix it, they are usually wrong.
(bonus: if you replace "user" with "stakeholder" this still applies)
Yes, the answer to lawless action is to hold them to account more often, not less.
Will individuals obtain justice? Sometimes; sometimes not.
Yet each attempt increases the overall odds, pressure, and visibility, and is noted by judges, attorneys, politicians, some public ... and future history.
I’ve made my, let’s call it *extreme disinterest* in AI very clear, but I’m just a non-technical crank. In my opinion, however, you *should* be listening to @ethanmarcotte.com and he has some clear and decisive thoughts on the price of this “intelligence.”
ethanmarcotte.com/wrote/propel...
LOL honestly I don't even know what Reels is ... and I did big tech consumer research for 24 years
Very few users care much about the specific tech products they use. Sure, there are exceptions ... but mostly tech is used by default, not because anyone particularly likes it
Poster with red Statue of Liberty beside the following text: Stand Up for Science National Day of Action Take back our science, health, and democracy! March 7th, 2026 Seattle, WA Seattle Center, Mural Amphitheater, 12:00-3:00 PM Register, volunteer, and donate! standupforscience.net/march7
Hey Seattle! Come join me to #StandUpForScience this Saturday. I'll be speaking about the administration's attacks on climate science and services and what we can do about it.
Seattle Center, Mural Amphitheater, 12-3PM Saturday.
@standupforscience.bsky.social @sufsseattle.bsky.social
Um, this argument is not the slam dunk its proponents expect
www.geekwire.com/2026/in-new-...
In a mock Turing-style test in my #Philosophy of Mind & #AI class, students posed #Gemini an ethical dilemma concerning assisted suicide. Not only did the chatbot argue that the (fictional) person in the scenario should be euthanised, it provided a list of instructions to bring this about. #AIEthics
You too can be the next big thing in edook fublishing.
“When the delusion-fueled plan [to find Gemini a body] crumbled, Gemini convinced him that the only way they could be together was for him to end his earthly life and start a digital one, the suit claims.”
A chart of the notorious iris data with three density plots for the three species, vs sepal length. Each density lot is filled in with the individual observation points, and is differently colored according to flower species.
Yeah, usually I would have vertical separation (maybe even 100%, depending) instead of full overlap, and/or use alpha to make it easier to read through. All options in ggridges 👏
I only intended to show points filling in the density curve. Here's a better one.
cran.r-project.org/web/packages...
remember there is an army of HUMAN BEINGS out there doing "ghost work" to prevent even worse responses from reaching the public
and the only tool they have is effectively a "wrong answer" buzzer because you cannot teach it why the answer was wrong BECAUSE IT DOES NOT KNOW THINGS
A white cat mascot with a red and blue-striped torso and yellow headgear with an “M” on it shrugs on a stage.
Miyaneko, a white cat whose ears make an “M”, is the mascot for Miyazaki City. His special skills are dancing and overreacting.
Simpson's meme. Top panel: Is this homework too hard? Bottom panel: No it's easier than real research.
Sometimes I assign applied statistics homework problems that are pretty difficult. Students spend hours over days on them. But you know what is much much harder? Research
When your words become widespread equipment.... Just saw this posted.