domain:eigenrobot.substack.com
Tree
2hr ago
That’s the definition of an analogy. You did say the female body is one of the most valuable possessions on earth. Your thesis is that her high value makes her insecure. I think it is the gulf between her appraisal of her own value, which is externally reinforced, and her actual, lower, value.
self_made_human
Grippy socks, grippy box
2hr ago
Excellent work as usual Dase. I was sorely tempted to write a K2 post, but I knew you could do it better.
challenges the strongest Western models, including reasoners, on some unexpected soft metrics, such as topping EQ-bench and creative writing evals (corroborated here)
I haven't asked it to write something entirely novel, but I have my own shoddy vibes-benchmark. It usually involves taking a chapter from my novel and asking it to imagine it in a style from a different author I like. It's good, but Gemini 2.5 Pro is better at that targeted task, and I've done this dozens of times.
Its writing is terse, dense, virtually devoid of sycophancy and recognizable LLM slop.
Alas, it is fond of the ol' em-dash, but which model isn't. I agree that sycophancy is minimal, and in my opinion, the model is deeply cynical in a manner not seen in any other. I'd almost say it's Russian in outlook. I would have bet money on "this is a model Dase will like".
Meta's AI failure are past comical, and into farce. I've heard that they tried to buy-out Thinking Machines and SSI for billions, but were turned down. Murati is a questionable founder, but I suppose if any stealth startup can speed away underwater towards ASI, it's going to be one run by Ilya. Even then, I'd bet against it succeeding.
I don't know if it's intentional, but it's possible that Zuck's profligity and willingness to throw around megabucks will starve competitors of talent, but I doubt the kind of researcher and engineers at DS or Moonshot would have been a priori deemed worthy.
orthoxerox
If you can read this, you're using a custom theme
2hr ago
pak chooie
Does it have to be a Tesla?
pak chooie
MaiqTheTrue
Renrijra Krin
2hr ago
I mean if you’re doing a female centric hobby and your video content is mostly watched by other women, you might be able to get by with doing that, but even “disembodied hands” videoed will read “woman! Who happens to do X hobby,” when the audience contains more men. Even their voice over the internet, or a chosen screen name in gaming and they become a Woman and thus get treated like an object of desire rather than “just another dude playing an online game.”
orthoxerox
If you can read this, you're using a custom theme
2hr ago
If it's the AI thread, what do you think about diffusion models for text?
self_made_human
Grippy socks, grippy box
2hr ago
I don't think anyone nominated me for a UVP, so I haven't had the opportunity. I probably would nominate you if it came up.
(Maybe you're thinking about the doge contest)
DaseindustriesLtd
late version of a small language model
2hr ago
There probably isn't a delay, plans to ship it in May to capitalize on the hype were entirely hallucinated by jo*rnalists as far as I can tell. It might take many months yet.
Thank you. I will clarify that by RL, I don't mean bog-standard RLHF, but more recent techniques like RLVR that have been around since o1.
BurdensomeCount
—Your em dash is showing...
2hr ago
I'm just waiting for DeepSeek R2. Not happy with the delay and while R1-05-28 is pretty damn good it isn't at the very top. K2 is non-thinking which means that while an excellent base model it isn't the best of the best when quality rather than speed matters.
DaseindustriesLtd
late version of a small language model
2hr ago
It's not really verbose in normal use, rather the opposite. It is verbose in agentic mode, writing docs and commentary rather than thinking traces. RL has been used for all serious LLMs since GPT 3-instruct, this is independent of the current long-CoT paradigm. It is dubious that Kimi has been trained on CoTs because it doesn't do them. More likely, its training data is largely final outputs of a reasoner (like Kimi's own 1.5/1.6). They have a section in the paper on 1.5 about penalizing verbosity.
2rafa
2hr ago
Great post, you’re one of my favorite commenters here. This makes me wonder if I ever did the User Viewpoint series. I think I did (maybe @self_made_human nominated me), but I can’t remember.
2rafa
3hr ago
The NYT’s house style was deliberately constructed in opposition to the British style of quoting everything.
NYT: Earthquake kills thousands
[London] Times: Earthquake “Kills Thousands”.
self_made_human
Grippy socks, grippy box
3hr ago
yes yes another post about AI, sorry about that
Feel that AGI baby!
It's obvious what the trends are. I predict that, on the midnight before ASI, the Motte's going to be 124% AI commentary. It might even be AI doing the commentary.
It's a primarily agentic non-reasoner
I have read claims that it's a pseudo-reasoner, and it was trained on COT traces and had RL done even if it doesn't use explicit reasoning tokens itself. I've also heard that it's 3x as verbose as most NRLLMs, almost on par with RLMMs, making the distinction academic. This was on Twitter, and I don't have links handy. I'm not sure how strongly to index on that.
self_made_human
Grippy socks, grippy box
3hr ago
A lot of the grognards over on HN don't think it counts, but they're the type who wouldn't accept blowjobs in heaven if it wasn't using a standard Apache license.
Hahaha yes, YES! Finally I can dust off my SICKOS shirt. Your move, Anthropic.
Factually, I concur with posters below that actually this isn't markedly different, and in some ways more basic, than the already possible setups for AI gooning (if the system prompt posted on X is real). Not a big fan of the voice either tbqh, although the model rigging seems good, and manic pixie Amane-Misa-at-home is a more natural fit for an AI waifu at a casual glance than whatever eldritch horrors Replika has. I'll likewise point out that while this incarnation is obviously aimed at men, the potential for female appeal is much bigger, and is in fact only a model/voice swap and a system prompt away. Not sure who is actually going to pay $300/mo for it however, the porn equivalent is literally free, and an equivalent ai gf via OR or something is much cheaper if a hassle to set up. Normies hate trivial inconveniences almost as much as steep price tags, I don't think this is how you get buy-in from them, but I assume this is literally the first step and we'll see further developments.
Regardless of technical chops, the real value here is of course exposure, a first decent shot at normiefying the whole edifice. Elon may be a fake gamer, the gravest insult I can levy against my fellow man, but fringe interests make for strange bedfellows, and I'm glad to see the first public attempts at rather literal waifutech make the twitterati seethe. I know what I'm following for the next few days. Though I agree with the OP that the whole Mecha-Hitler brouhaha was 100% a publicity psyop in hindsight, the timing and subject matter is too conspicuous. Based?
On another note, I think that people invoking the Torment Nexus are kind of missing the point. I don't think "authentic" humans are threatened (NB: screeching xitter denizens not considered human authentic). Even the most literal goonette femcel I know consistently finds much more joy in talking/RPing with an actual human over a chatbot, by a rather wide margin, even if the chatbot wins out on 24/7 availability.
Instead, I think the real horror potential here is - may Allah forgive me for uttering this word - gacha games, or more broadly chatbot integration and monetization. I've recently gotten into Girls Frontline 2, and seeing the manic pixie grok gf clicked something together in my head. I can already see the framework, the jewish tricks are practically manifesting before my eyes: gacha girls have affinity/bond levels (here increased by gifts = in-game loot), a certain level of bond unlocks the chatbot functionality with the given waifu, further levels relax the guardrails or change the system prompt (reflecting increased affection)... you get the gist. My cai/Chub experience tells me gacha girls are some of the most popular interlocutors anyway, so the match is eminently natural.
From there the potential for added deviltry is almost limitless:
- obviously, 3d model visibly reacting to prompts like not-Amane-Misa here
- outfits for said 3d model, those are already a gacha staple but maybe some kind of added animations or different "backgrounds"/environments for the chatbot part? (i.e some hypothetical maid costume, with some abstract cafe setting written into the chatbot's system prompt if the costume is on?)
- limit the amount of prompts per day (vantablack ninth circle hell tier: offer paid refreshes)
- lock explicit NSFW behind a paid item (e.g. GFL2 has a
marriageCovenant mechanic, the ring is single-use and costs $5) - give the waifus some kind of actual stat boosts for "cheering them up" before gameplay, grading incoming user prompts to this end like Grok seemingly does (I eagerly await the inevitable rizz meta guides on how to best word your prompts for the optimal boost)
- some kind of voice command integration built on top as an afterthought? GFL2 is turn-based xcom-lite so I imagine voice commands can work given an autistic enough framework under the hood
Granted I sense the danger of metaphysical cuckoldry Chevrolet-tier oopsies and general bad press here, a man in pursuit of his coom is endlessly inventive, but as long as the chatbot is sufficiently insulated on the backend (also just imposing harsh character limits already neuters most prompt engineering) I think this can work. Though it probably won't be a Chinese gacha given the dangerously free-form nature of chatbots, and I don't think anyone else is deranged bold enough to try.
self_made_human
Grippy socks, grippy box
3hr ago
Other than reach and better animation, I don't think this is different from the AI companions that have been available for a while. Replika, the most famous one, will already do NSFW ERP. And yeah, there are men (and women!) who have decided their Replikas are preferable to real people.
That fact that it's animated is a big deal! Men are visual creatures, and the fact that previous ERP was textual made it far less appealing to the average dude, if not woman. Of course, jerking off to anime tiddies is still not a preference of the majority, but it's easy to upgrade to photorealism. That'll get more people.
I predicted this outcome ages ago, though I'd have said it was inevitable and obvious to anyone who cared. It's priced in for me, and I agree that it likely won't be catastrophic.
self_made_human
Grippy socks, grippy box
3hr ago
I don't doubt that, but once again, that doesn't mean that the vast majority of people are receiving any actual attention from the CIA.
RenOS
something is wrong
3hr ago
Zooming out a bit, the UK has currently similar problems (exploding general healthcare/welfare costs leading to low-priority care being sidelined) despite an establishment politician winning, germany has similar problems despite an establishment politician winning, but somehow in the US it's Trump's fault. Even the specifics - rural hospitals getting disproportionally closed - are the same. Remember, healthcare cost graphs over time look like this, and this is % of GDP, so it's even more crazy in absolute terms. That growth is not sustainable, and indeed is starting to not getting sustained much longer. That doesn't mean Tump's cuts aren't higher than elsewhere nor that they have no negative effects, but the framing here is quite questionable.
self_made_human
Grippy socks, grippy box
3hr ago
My apologies. I was thinking of this related thread, and it's not you I was arguing with.
(Some might even call the mistake I made a hallucination, hmm)
ace
3hr ago
The sats observe everything and images are kept forever. You are just picked out and individually tracked by software after the fact.
(I don't know it for a fact; it's just my best guess from extrapolating from leaks. I'd bet money on it if there were ever some way to get at the ground truth.)
So, some observations. First, sorry dude, but I have major side-eye for your ability to evaluate literary quality. :p
You hit below the belt. Reverend Insanity is Peak Fiction and I'm going to go down swinging!
As you probably know, even the most powerful LLMs do not have a context window large enough to store an entire large novel in memory, let alone a series, and you can't directly upload embeddings to GPT or Claude
1 million tokens is a lot! (Gemini 2.0 had 2 million, but good luck getting it to function properly when it's that full). That is 750k words. All of Harry Potter is just over a million.
I'm going to ignore Llama here, since even if it has a max 10 million token CW, mental retardation is not improved by the fact that there's a lot more of it. And why shouldn't I? Even Zuck has chosen to forget that particular failure.
I've uploaded whole medical textbooks into them without major issue. Not tiny books either.
As long as you can keep it on track, I have found that some of the GPT and Anthropic models are... not terrible as beta readers. They point out some real flaws and in a very generic sense have an "understanding" of pacing and tone and where a scene is missing something.
I am most personally familiar with uploading chapters (often half a dozen) of my own work, which works well. If I was less lazy, I'd probably be saving summaries of the whole thing and stringing them together. (Royal Road makes it so you can't export an epub of your own fic without paying, and without that option, I'd be doing a lot of copying and pasting)
When asked for critique, some of the issues raised were cogent. Too much jargon, uneven pacing and so on.
Some of that was intentional, such as the fact that since the excerpts were lifted from a larger work, most of the jargon was previously explained at one point or the other. I also have no shame about making potential readers resort to keeping a Wikipedia tab open on the side, it's niche hard scifi and I want to flex. Other issues are well worth amending before publication.
I haven't had the good fortune of having very many professional authors or editors review and critique, and I don't doubt that they'd probably give me even more useful feedback. Yet what I get is quite good and elevates the final product!
I still think we'll need true AGI to write an actual good novel. When you show me an AI that can write a coherent series, with multi-volume character arcs, plot seeds planted in early books that clearly pay off in later ones, literary allusions and metaphors that aren't just clumsy pulled-off-the-shelf ones but deeply enmeshed in the story, and a recognizable differentiable style (in the same way that fans can read Dickens or McCarthy or Hemingway and immediately recognize the author), I will believe we're there.
That aligns well with my own stance. A large novel is an unwieldy thing, let alone a good one. We're still at the competent novella or subpar novel stage, but I must stress that's a comparison against the very few human authors who make big bucks and/or accrue critical acclaim. Most things humans or LLM novelists write are slop, the former just don't scale as hard.
DaseindustriesLtd
late version of a small language model
3hr ago
Relevant from Lambert: The American DeepSeek Project
While America has the best AI models in Gemini, Claude, o3, etc. and the best infrastructure with Nvidia it’s rapidly losing its influence over the future directions of AI that unfold in the open-source and academic communities. Chinese organizations are releasing the most notable open models and datasets across all modalities, from text to robotics or video, and at the same time it’s common for researchers worldwide to read far more new research papers from Chinese organizations rather than their Western counterparts.
This balance of power has been shifting rapidly in the last 12 months and reflects shifting, structural advantages that Chinese companies have with open-source AI — China has more AI researchers, data1, and an open-source default.
[…] The goal for my next few years of work is what I’m calling The American DeepSeek Project — a fully open-source model at the scale and performance of current (publicly available) frontier models, within 2 years.2 A fully open model, as opposed to just an “open weights” model, comes with data, training code, logs, and decision making — on top of the weights to run inference — in order to distribute the knowledge and access for how to train AI models fully.
etc. He overstates the cause, perhaps. America doesn't need these egghead communist values of openness and reproducibility, the free market will align incentives, and charity shouldn't go too far. But he's pointing to the very real fact that China, and not on the state level but on the level of individual small companies with suitable culture, is the only country bringing transformative AI not locked on corporate clusters closer to reality.
I don't think the Birkenhead drill only applies if the women in question aren't barren. Of course the value bestowed upon women is ultimately an evolutionary adaptation to the reality that only women can bear children. But in practice, even barren women are still seen as Wonderful™ in a way that NEET men aren't.
self_made_human
Grippy socks, grippy box
3hr ago
"Watching here doesn't mean something so casual as the fact that there's a sat that Incidentally oversees my geographical location from gestationary orbit.
Us psychiatrists might be nerdy and out of date, but we're not that far gone, and this would be discussed before committing someone"
The fact it can, in "some cases" be true, makes it a non-bizarre delusion. The quote specifically says "extraordinarily unlikely", and I'd probably take some international arms dealer who told me so more seriously.
Does that mean the Maya were right after all?
More options
Context Copy link