Matt Brunig is a socialist lawyer who loves using AI. He might even be a classic example of a lefty among his political allies who takes a contrarian position on AI. Earlier this year, in an article titled “The Left Is Missing AI,” he told Transformers’ Dan Cagan-Kuns that he was concerned that “the left is refraining from using the tools when the right is not using them.”
But at the time, it was less clear to me how it would follow that left-wing goals such as forming workplace unions, building working-class movements, and electing left-wing politicians should benefit from AI.
I can’t say that yet abundantly That’s obvious, but Bruenig certainly made an interesting YouTube video about using Claude Code to the left. He’s a policy geek, so this video isn’t for those who don’t feel comfortable watching a 51-minute video on statistical weighting, but if you can get over that hurdle, I recommend the following:
This comes after a recent Wall Street Journal article about young people living with their parents. The idea was framed as an almost natural means of frugality, rather than the hardship it once was, a strategy smart young people are employing, perhaps to reunite after a breakup, or to “save money for a new home.” Earlier this month, I heard drive-time DJs talk about this viral article on the radio. Part of it was about the undeniable fact that housing is too expensive in America, but the part about cultural change didn’t seem accurate.
But this key statistic seemed undeniable.
“49% of adults under 30 said they lived with their parents last year, an increase of 12 percentage points from 2019, according to the Federal Reserve’s latest Survey on Household Economics and Decision Making.”
The 12 percentage point increase is the first in six years and reflects not only the shock of the COVID-19 era, but also a sudden and widespread shift in attitudes.
And according to Brunig, with the help of the Claude Code, that’s simply a bad conclusion drawn from the statistical garbage of the Survey of Household Economics and Decision Making (SHED) survey data.
In the video, you can see that Bruenig has already loaded the survey data into several folders (inside Yazi) and is using Claude Code (equipped with the Ghostty terminal emulator and running tmux to create panes in Ghostty). In a previous video, he showed how to use this setup to perform microdata analysis.
And while I’m no statistics expert, I’d be hesitant to simply agree that the Journal’s conclusions are wrong based on some YouTube video, but it’s impressive to see Bruenig basically prompt Claude by verbally saying that the statistics are a bit questionable, and then have Claude code go into Bloodhound mode and quickly jump through what seems to be pretty obviously bad math.
The illusion of a sudden 12% increase is reportedly caused by a “surveying artifact.” One year’s study found that the 18- to 29-year-old tranche of SHED had too few teens who were more likely to live with their parents, but another study found that the tranche was more balanced in age. On the graph, the results sampled this way look like there is a large increase in the 18-29 age group living with their parents, but this is an outlier caused by statistical noise, and the actual recent increase is probably about 3%.
This probably means that there hasn’t really been a big attitude change and that people are probably just stuck living with their parents. It’s the same as before.
It would be great to use AI to clearly prove this point with minimal effort. Does that mean the left should wholeheartedly embrace AI? That is a discussion for another time. But assuming that the mathematics can be independently verified, this is an example of AI sniffing out truth, and it might at least be of value to someone desperate for a left-wing proof of the objective truth of, say, a maxim of historical materialism. under consideration Things like Bruenig’s AI tools as helpers. or not. Please don’t get angry.
