Will Harrison

Super cool finding

i think enthusiastic LLM use is mostly a stack of cognitive biases, unacknowledged plagiarism, and unmet needs in a trenchcoat

but also my main objections aren't about them being bad at tasks so i don't care if you think they've gotten better at it

Skateboard stop-motion

To add to this: now is the time we need human to human training and transfer of verifiable research and scholarship skills.

As much as tech bros want to replace your jobs, legitimate scientists are now more important than ever.

/sarcasm

I for one am glad we now pay for these mind blowing AI features.

Neat. The details will matter. Got a reference I can look at?

Presumably one per minute of operation?

"we just need more training data bro"

It still remains the case that most "AI" functionality is being pushed by multibillion dollar companies with little or no evidence for the claims, while scientists who work in the area continually call bullshit and point to simple logic and foundational psychology.

And formally provably so because to drive you need theory of mind. You can indeed show with maths the limits of maths since it's a formal engineered system with knowable limits, see Gödel etc. We created trains which ban humans from tracks so the system can be safely automated, not so for roads. 1/n

"Tech killed online education" is a heck of an outcome.

Yeh I don't doubt some of the patterns are obviously dubious, but it could still be useful to put a number on it, even if it's "We couldn't find a real human who had this pattern of data in any of our N lab participants."

I learned so much about frequentist vs Bayesian frameworks from the many debates on the old Twitter. I’m also sure some people in my real life science circles started as twitter contacts.

Possibly in the OSF repo, but it'd be useful to know from a lab-based experiment with real humans what is the expected frequency of each of those "bot" patterns.

Even images of kiwi fruit have an overrepresentation of horizontal features ...........

Four panels showing how an image of a kiwi fruit can be filtered to understand contrast energy.

So @reubenrideaux.bsky.social and I decided to run an image-processing workshop at this year's EPC/APCV. We will be teaching people how to compute the contrast energy of kiwi fruit, I guess. Sign up now: visualneuroscience.auckland.ac.nz/epc-apcv-2026/

@expsyanz.bsky.social

The Rise of Eyes Began With Just One

Vision Science makes the front page of NYT!!
featuring work from @neurofishh.bsky.social and @denilsson.bsky.social with comments from Berkeley's own @karthikshekhar.bsky.social
#visionscience

www.nytimes.com/2026/02/23/s...

The utterance of "A.I.'s inevitability" is one of the most stark pure performatives I've seen in my time working in higher ed. Every time it is uttered, it is clearly not reporting a fact about the world but instead actively trying to create the reality it narrates. We can and must refuse.

Anat is not only a brilliant scientist but also one of the nicest and coolest people I've ever met.

It's AI

lol

I feel like this should be out there but I’m drawing a blank… @rebeccakwest.bsky.social has a paper that models errors for separate judgments per trial, but not a typical dual task paradigm. I’m wondering if she may have any leads?

I tried to take a photo of a grasshopper on my windshield, but now it looks like a gigantic bug destroying the town.

Sounds super cool! I'm very interested in the fancy new tools, but I am still skeptical about how well they address many of the questions I personally am interested in. You may want to read more about me here: willjharrison.github.io#about

Thanks for nothing, bsky!

Anyone want to do a PhD with me at the Sunny Coast? I'm recruiting, and I wanna do some fun psychophysics (but the possibilities for the PhD are very broad). Domestic students only, sadly.

In case y'all happen to know someone:
@nataliepeluso.com
@reubenrideaux.bsky.social
@visnerd.bsky.social

Bluesky Map Interactive map of 3.4 million Bluesky users, visualised by their follower pattern.

I made a map of 3.4 million Bluesky users - see if you can find yourself!

bluesky-map.theo.io

I've seen some similar projects, but IMO this seems to better capture some of the fine-grained detail

Our uni also has an elite athletics program with a few Olympians, but I hate that there is policy that grants them specific allowances for their study. Personally, I'd prefer it if ALL students could get accommodations for non-academic stuff, or none of them...

We are all biased estimators, and our estimates of how biased are also biased.

It's bias all the way down.

Great study and clear results. Did you do any exploratory analyses at the individual level to see if there are some individuals whose pupil correlates with vividness?