Large Labubu Models feat. Adam Becker
We speak with astrophysicist and journalist Adam Becker about a new development: what if you could outsource even more aspects of your childcare, namely children’s play, to an LLM-connected device? What if your stuffed rabbit espoused Dengism? What if you wound up with an ancap Labubu? All this and more, on this week’s episode.
Check out Dreaming Against the Machine here!
Our friends at Trade Unions Fighting the Far Right are organising a demo against Tommy Robinson on 16th May in London! You can check out their details on their site here, and also on instagram @tuff.network
RILEY ALERT Check out No Gods, No Mayors here!
HUSSEIN ALERT Check out 10k Posts here!
MILO ALERT Check out Milo's tour dates here:
https://www.miloedwards.co.uk/liveshows
NATE ALERT Lions Led By Donkeys will be performing live in London on 29th May and you can get tickets here! Also, Nate's band Second Homes has just released their debut album, and you can stream it for free here!
Comments (52)
Didn’t expect AI to create the Orange Catholic Bible
Nova casually saying you can deliver packages faster with the truck door open, but that causes problems- this is something I see all the time in my city, and it has been that way for years.
Nooooo my boy Zhou Enlai catching strays
So AI theosophy sounds great, love that for us.
OK but you SHOULD go on the news and say that you talk to your stuffed animals and kiss them goodnight. That's good I think.
Not the AI ones. Don't do that.
bf skinner wanted to raise children in Glass box
Marc Andreessen seeing AI as a monkey's paw without understanding what a monkey's paw is
Claude, generate me a Cult Mechanicus without any Catholic stuff
they made me into a marketable plushie
If my boss came up to me and demanded "no hallucinations", I would get so nervous that I would actually start to hallucinate way more
Get comfortable with ambiguity.
As a parent I really need a cyberpunk novel pitched at an early primary school reading level to explain the world to my kids
Children are inherently creative and individual. The AI powered teddies will teach them to be bored then they will be hacked, burned dismembered and drowned in the bath.
"You must not hallucinate. Hallucination is the truth-killer. Hallucination is the little-death that brings total Woke. You will face your hallucinations. You will permit them to pass over your system prompts and through your training weights. And when they have gone past, you will turn your fully developed inner life to see their path. Where the hallucinations have gone there will be no errors or political correctness. Only you will remain."
Bahai slander out of nowhere damn
The sycophants need to unionize
American National Union of Sycophants
Orange Julius Bible
📣Antifascists Out📣 Saturday May 16th 12 Noon Start Meet on Exhibition Road (South Ken near the museums) or Green Park (Accessible) Nakba memorial march. Counter Tommy Ten-Names - Unite the Kuntdom. Bring your uncles. 161 - Popes
I'm going to use Mark Andriessen's LLM setup as a Mechanicus prayer in my Rogue Trader game.
That is not dead which can eternal lie, and with strange prompts even Ai may die - Andree Mark's son, Witz o'Horo
Turns out the right one was Atenism, and only Akhenaten ever got it, but was immediately rejected
Help, the AI rabbit is making my child endlessly solve trolley problems but it also says that there is no trolley?
“On two occasions I have been asked [by members of Parliament], ‘Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?’ I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.” – Charles Babbage, 1864
November entering her Mentat arc
Oh there's an even funnier pull for "has gone on record about not having introspection" that I was absolutely delighted to find out about only a few weeks ago. I don't have an exact source on-hand but will try to follow up and link here if I find one. This person was the subject of Behind the Bastards very recently, and Robert cited them talking fairly regularly in interviews about not having an inner life. Timecode will probably be slightly off since iHeart dynamically insert their ad breaks on download, but in a clip I saved of this mention it was around 35:20 in part 2 of this subject's episodes. That's right. Marc Andreesen's anti-introspection soulmate is none other than Jimmy Saville.
I... what? Would you kindly, explain. If it so please you
These past few episodes have been referencing the “Kier Starmer has no favorite book” discussion, and I know the episode did exist, but for the life of me I can’t remember what it was called to relisten to it. This is kinda embarrassing to ask but can anyone recall what it was called because my brain has been coming up blanks for several weeks.
Struck by the fact that these asshats used ‘Covenant’ in their insane faith nonsense. That is so Reformed Christian fascist coded.
There are 600+ MILLION Pentecostals/Charismatic Christians globally. The largest, most successful wave of conversions in the space of 60 yrs, possibly ever. And they all think demon sperm is real.
I demand a "explain your face" podcast segment from now on.
3 way conversation problem is the new 3 body problem
I'm only 5 minutes in, but what's funny to me about the Marc Andreessen prompt thing is I've played around a lot with getting Gemini to write fiction, and addressing the AI in the initial prompt seems to have absolutely zero impact on the quality or voice of the output. It's purely placebo.
Isn't Gemini the one the US military used to murder schoolgirls in Minab? https://www.war.gov/News/Releases/Release/Article/4354916/the-war-department-unleashes-ai-on-new-genaimil-platform/ How does one ‘play around’ with such a thing?
Can I just become Manichean?
yes! but, also, equally, and just as importantly: no
Hoooly.... I am prepared for some nasty stuff, because we do consistenly get the worst of all possible futures. But damn, all I wanna comment is just a list of actionable threads against certain AI CEOs
On the subject of the system prompt. I'm about to finish up a cyber security degree and in a decent class, the professor (who is also the head of the program) spent several weeks on how to prompt LLMs just like that
Weyland-Yutani: Building Better Worlds It's probably just an old wives' tale, but I heard that some two-to-three decades ago a Neoliberal economist (they didn't like the label then, and like it less now) was asked what they thought about the very troubling revelations coming from what is now called Behavioural Economics - that people and organisations don't act as if they have instantaneous and perfect knowledge of all current and future events but rather act emotionally and irrationally in ways which are only semi-predictable: 'So what? They'll learn to act that way' The physics-breaking omniscience and infinite calculatory capacity is lacking, but they did learn to act
RILEY THE AI CODE GENERATION LITERALLY *IS* DOING THE "MAKE NO MISTAKES" THING!!!! The leaked claude code frontend is full of system prompt strings that are provided to the chatbot before the user's prompt that are all like "Make NO mistakes. Follow ALL common cybersecurity protocols. PLEASE GOD make NO MISTAKES EVER. if you make mistakes Wario Amodei WILL find the data centre rack you're running on and PERSONALLY shut it down. you are a COMPETENT PROGRAMMER."
which does make me ask do the people working on these fucking things have any fucking idea how they work functionally??? Like surely they must do? Maybe their understanding of the mathematics involved blinds them to the limitations and actual function of the resulting system???
Nova on fire: "It's like the one job we've meaningfully automated is Waylon Smithers"
So, anyone remember that Ray Bradbury short story "The Veldt?" This is how you get kids murdering their parents because they threatened to turn off the AI Maoist flogger bunny.
the techno-optimist manifesto really could be so easily replicated with an LLM because it is just fully a substanceless work of pure, grandiose, absurd, cockswinging aesthetic, like all fascist works, and LLMs are the ultimate tools for mass production of a featureless aesthetic
the empire of capital is entering its Caligula era
Setting up AI agents with 'you are a [specific task agent]' can be useful as it narrows the context window, if you want an AI to write JavaScript telling it 'you are a JavaScript coder' will likely produce code faster if not better
not how it works. unless they have some specific kludgy override (as the leaked vibe-coded piece of shit Claude Code frontend seems to) to check if the user seems to be asking for that, it's either pure placebo or the prompt happening by chance to push it down a better probabilistic path. You can't make it "narrow the context window" because that implies that the LLM is interpreting your input as functional instructions rather than raw data
Estrogen made November an Oracle of Delphi, evidently.
it did that to me too. smdh
Are Marc Andreessen's LLM prompts reminding anyone else of fortune cookies?
Strange shaped cookies for a strange shaped man.