Pixels and Patterns
Subscribe
Sign in
Home
Notes
Archive
About
Online survey panel quality in the age of AI: Some reflections from recent surveys
Lessons on AI use and online sample quality from four countries
Oct 15
•
Steven Denney
2
2
3
JEV Is Fast. Is It Better than LLMs?
Accuracy, cost, speed and calibration across seven text-as-data replications.
Oct 7
•
Steven Denney
and
Matthew DiGiuseppe
4
1
July 2026
Talk to your terminal
Dictating prompts to a CLI agent beats typing them.
Jul 13
•
Steven Denney
4
1
Who are the Americans?
On the 250th birthday of the United States, I reflect on thirty years of survey data to show a national identity in flux and make a bullish case for…
Jul 4
•
Steven Denney
7
3
June 2026
From hallucination to verification
AI has made fabricated citations commonplace. Used correctly, AI can make them rare. I review the problem and offer some "skills" as part of the…
Jun 30
•
Steven Denney
14
2
7
Vision-language models are weirdly good at Old-Korean
I stress test VLMs on Old-Korean and it works surprisingly well.
Jun 26
•
Aron van de Pol
1
A council of models
Exploring how we can use LLMs like a team for empirical research.
Jun 26
•
Steven Denney
8
3
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts
