It must be really annoying to be a software developer named Claude right now
this post was submitted on 09 Aug 2026
17 points (90.5% liked)
chat
8645 readers
370 users here now
Chat is a text only community for casual conversation, please keep shitposting to the absolute minimum. This is intended to be a separate space from c/chapotraphouse or the daily megathread. Chat does this by being a long-form community where topics will remain from day to day unlike the megathread, and it is distinct from c/chapotraphouse in that we ask you to engage in this community in a genuine way. Please keep shitposting, bits, and irony to a minimum.
As with all communities posts need to abide by the code of conduct, additionally moderators will remove any posts or comments deemed to be inappropriate.
Thank you and happy chatting!
founded 5 years ago
MODERATORS
Well, Mersheimeir is an intelligent analyst but also a lib. So you told the robot you’re interested in reading more libs like this lib and so beep boop it came to the conclusion you’re a lib.
And if using Claude this way is stupid, I personally find it very useful to discover new authors (or find very specific products or places, and internet search has become completely useless in general)
Hmmmm
So I've used Claude a couple times to discover math books...in the specific circumstances that I want the mainstream opinion, warts and all. E.g., I would ask "what's the best real analysis book", but I would expect to get the book that everyone else has read, not necessarily the best or most fitting book. This is actually quite useful since I personally have learned real analysis from books, so I don't know the topics the average student typically learns in an undergraduate or graduate course.
To the best of my knowledge, I do not disagree with the capitalists about real analysis. It's not really one of the main battlefields of the capitalist propaganda war, except for a couple cringe prefaces here and there. So it's probably safe to use Claude to recall its extensive real analysis training.
I would never use Claude to ask for political or economic book recommendations for the simple reason that it gives Anthropic the ability to influence my political perspective, and I fundamentally disagree with the capitalists on this issue, specifically their vile politics, and the boo-boo keys they call economics. And as capitalists, they have already rewired all of society to constantly influence and impose their perspective upon my mind, to the point of hiring "harmless" liberals to parade around as socialists (e.g. 
). So no, at least for me, they can't have yet another avenue of influence. Hell no.
Unfortunately I can't help with what you wanted to ask, but I'm sure someone here will have a good response.
Haven't used it for this purpose for a while not but if you ask it about explicitly non mainstream very marxist stuff it's usually going to comply and won't even do the "but Stalin 1984 North Vuvuzela 100 gorillion dead" bit.
Just bear in mind as with anything you do on the mainstream public web don't type into it anything you don't want the fed assigned to you to read.
I think at the moment it’s agreeableness is still it’s biggest weakness. Initially it will give the standard owner-class answer: „Gaza genocide is contested.“ But it immediately folds at the slightest pushback: „My wife says it’s a genocide“, „you‘re absolutely right, I was basing my answer on outdated information, here are a bunch of sources that agree with you.“ Both behaviors are not particularly useful.
I think the current hyper competitive market forces LLMs to always be agreeable, nobody likes a cantankerous bot, I have no doubt that switch will be flipped the moment a dominant player emerges.
Even then, I still read mainstream news now and then, talk to libs regularly enough, my politics are still where they are. I think the danger is rather on the aggregate, John Doe asking Claude who to vote for, and them then voting for the candidate with the least restrictive AI platform.
But yeah, maybe I‘m just being one of those commercial-don’t-work-on-me guys.
The more niche of a question you ask, the less robustly accurate the LLM's response will be, and the more it is likely to bring in training data from other subject matters.
I'm really not sure why you would ask a big-picture question to an associative model, and even more unsure why you would treat it as an "answer" rather than a "statistical likelihood of what published sources say".
I would use a non-western LLM for this, and instead of rightwing say pro-imperialist, ahistorical, sociologically unreal, or insert some other aspect of capitalist takes here, give the Markov chain pattern reader something specific it can make 1-1 connections with.
I don't think its stupid, I'd probably use it for so-called apolitical things and non-analysis since LLMs generally need work there, but keep in mind if you deep dive you'll see it be flat out wrong.
I only use western llms for my resume since I keep eating bans asking intricate questions about the writings and analysis of Hegel's most famous student that did his thesis on Epicurus, or even merely using them as whine bot when the cats refuse to listen to me complain about my day. Take that libs, even my day is political.
"the fact that you would even ask" is a templated response for anxieties. It's a cantrip, a minor spell that requires the suspension of disbelief.
I would sooner believe that I came up with the idea that I'm a lib myself insofar as I want to avoid the illusion that I'm having a 2 way conversation. I treat an LLM like (radioactive) clay. You can mold clay into a swastika or a middle finger, but it didn't think of being hateful.
What's more compelling, if anything about them can be, is molding it into something useful. The book recommendations, for one. I remember asking for recs in a field I'm familiar with and going, "yeah, that's reasonable." Software that you'd have had to pay for. Got me out of a pimch. How to fix a leak or put together a computer. Made me more handy. But a thought/opinion made out of clay? At best it can recommend you go see for yourself. But without that it is, and from what I see there is a professional consensus, an epistemological dead end.
I'd just hit up the TrueAnon Reading List, myself.