Need to let loose a primal scream without collecting footnotes first? Have a sneer percolating in your system but not enough time/energy to make a whole post about it? Go forth and be mid: Welcome to the Stubsack, your first port of call for learning fresh Awful youāll near-instantly regret.
Any awful.systems sub may be subsneered in this subthread, techtakes or no.
If your sneer seems higher quality than you thought, feel free to cutānāpaste it into its own post ā thereās no quota for posting and the bar really isnāt that high.
The post Xitter web has spawned soo many āesotericā right wing freaks, but thereās no appropriate sneer-space for them. Iām talking redscare-ish, reality challenged āculture criticsā who write about everything but understand nothing. Iām talking about reply-guys who make the same 6 tweets about the same 3 subjects. Theyāre inescapable at this point, yet I donāt see them mocked (as much as they should be)
Like, there was one dude a while back who insisted that women couldnāt be surgeons because they didnāt believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I canāt escape them, I would love to sneer at them.
(Credit and/or blame to David Gerard for starting this.)
This hits differently over the recent news that ChatGPT encouraged and aided a teen suicide.
transcript
Kelsey Piper xhitted: Never thought Iād become a ātake you relationship problems to ChatGPTā person but when the 8yo and I have an argument it actually works really well to mutually agree on an account of events for Claude and the ask for its opinion
I think she considers the AIs far more knowledgeable than me about reasonable human behavior so if I say something thatās no reason to think itās true but if Claude says it then it at least merits serious consideration
When an 8 year old thinks an AI is āfar more knowledgeable than me about reasonable human behaviorā that could lead a person to self-reflection. Could.
@nightsky @Architeuthis Computer programs are generally written to be basically polite for customer service purposes.
A generative text program that opened chats with āUgh, what do you want THIS time, loser?ā would likely find itself ā¦correctedā¦ by the parent company on short order.
@nightsky @Architeuthis the kidās not wrong, though š
In 12 years weāll get a book āMy Mom Outsourced Raising Me to AI and it Broke Meā
Found the original Tweet, with a reply from Yud himself:
Yes dude, thatās the main thing you should be concerned about of course. AI tools couldnāt possibly be bad in and of themselves, it has to be human tampering. Youāve always been very clear about that part.
tfw your gifted child syndrome resentment of adults is powerful enough to make you forget about your lifeās work
I was trying to figure out why he hadnāt turned this into an opportunity to lecture (or write a mini-fanfic) about giving more attack surface to the AGI to manipulate youā¦ I was stumped until I saw your comment. I think that is it, expressing his childhood distrust of authority trumps lecturing us on the AI-Godās manipulations.
and of course we all have root on the prompt, where - at will - we can just instantly impose all manner of will on the corporate vendor chatbot. yāknow, the chatbot operating in a service structured as much as possible to try to do what the corporate vendor wants to desperately maintain
(it continues to astound me that anyone takes yud seriously, at all, ever)
I feel dumber for having read that, and not in the intellectually humbled way.
I mean, tampering with the system prompt is definitely a kind of concern, given what weāve seen happen with Grokās tenure as mechahitler or Replika users finding their girlfriend no longer wanted them. But āmessing with system memoryā is the kind of sci-fi nonsense that should stay in a cyberpunk novel.
He has capital L Lawfulness concerns. About the parent and the child being asymmetrically skilled in context engineering. Which apparently is the main reason kids shouldnāt trust LLM output.
Him showing his ass with the memory comment is just a bonus.
āI have Lawfulness concernsā is just another way of saying āI deserve to be shoved into a lockerā.
System memory is just the marketing label for āhaving an LLM summarize a bunch of old conversations and shoving it into a hidden promptā. I agree that using that term is sneer-worthy.
Thanks for the clarification. I had definitely assumed that he meant some kind of God-AI-level attack that revolved around live editing the data or state in RAM or something.
That is his concern and not the billionaires behind it messing with the systems so much you cant prompt override it? Please tell me this guy doesnt work in AI alignment.
I have context that makes this even more cringe! āLawfulness concernsā refers to like, Dungeons and Dragons lawfulness. Specifically the concept of lawfulness developed in the Pathfinder fanfiction weāve previously discussed (the one with deliberately bad BDSM and eugenics). Like a proper Lawful Good Paladin of Iomedae wouldnāt put you in a position where you had to trust they hadnāt rigged the background prompt if you went to them for spiritual counseling. (Although a Lawful Evil cleric of Asmodeus totally would rig the promptā¦ Lawfulness as a measuring stick of ethics/morality is a terrible idea even accepting the premise of using Pathfinder fanfic to develop your sense of ethics.)
Not everything that Yud writes feels like it should be read in the voice of Augustus St. Cloud, but a lot of it sure does.
From elsewhere in the replies:
āTracingWoodgrains, who apparently feels that he is qualified to judge any other human being, ever, in spite of evidence to the contrary
KP writing a paper for the journal āNew Frontiers In Gaslighting Childrenā
āWhen the 8-year-old and I have an argument, it actually works really well to mutually agree on an account of eventsā¦ and then take cocaine together.ā
Acknowledging/validating each otherās feelings and finding a mutually-agreeable understanding of the conflict is already the hard part that most parents and kids arenāt willing to do. Talking to a chatbot after that just seems like you donāt understand the fucking point and are still trying to ābe rightā or whatever.