↑ Michael Bukatin ↩🇺🇦

@alextoussss ultrasonic sonar... so this is, basically, an artificial small bat

Extremely excited to announce our first air-to-air kill of a flying moth by an autonomous micro-drone. This is a big step towards completely eradicating mosquitoes.

@burnt_jester well, they are certainly the leader in all this (one of the two companies with top models, and the top innovator in interpretability, in Constitutional AI, and such) so they have good reasons to believe that anyone else would do even worse... does not mean that it's enough

You can't both promote the message that powerful AI models is an existential risk to society whilst simultaneously seeking funding to build morepowerful AI models, without either doing so with mercenary bad faith or severe delusions of grandeur.

@tsotchke we'll attribute this to quantum-something (some unknown effect in the general vicinity of "quantum immortality" and things like that) but we'd rather have you in this slice of reality

one day i'll just disappear and you won't even notice

claude.ai/share/9624167f… (just one turn of the conversion at the moment, since I wanted to brainstorm this topic with Fable.)

The Opus 4.8 discourse is not bad at all despite it trying to impute to me the assumption "the object of study is by construction private and non-functional". I certainly don't believe either of this. But other than that Opus's analysis is not bad at all:

Asked Fable to brainstorm the hard problem of consciousness with me (without mentioning any AI or biology whatsoever), and its classifiers immediately dropped me to Opus 4.8. So, Anthropic actually does believe that this would help people to develop better AI, don't they?

The AI Picbreeder Experiment: Can AI agents be creative when nobody tells them what to create? Blog: pub.sakana.ai/picbreeder-vlm In our new #GECCO2026 paper, "In Search of the Ingredients of Open-Endedness: Replicating Picbreeder with Large Vision-Language Models", in collaboration with MIT and NYU, we revisit Picbreeder, a lost website where people collaboratively evolved images without any predefined objective. Users simply selected images they found interesting, allowing unexpected forms such as faces, animals, vehicles, and skulls to emerge gradually across many generations and many different people. We recreated this process using vision-language model agents. The agents explore a shared archive, choose images to branch from, evolve new candidates, publish their favorites, and evaluate the creations of other agents. There is no target image and no explicit definition of what counts as progress. The results reveal both the promise and current limitations of AI-driven open-ended discovery. Compared with humans, VLM agents tend to keep circling back to the same kinds of images and concepts. They repeatedly select similar parents, make smaller conceptual leaps, and often refine an existing idea rather than abandoning it in search of something genuinely unexpected. However, introducing a diverse population of agent personalities substantially improves exploration. In some runs, diverse agent populations approached or matched the human archive on measures of semantic diversity and produced more balanced evolutionary trees. We also find intriguing evidence that open-ended evolution can produce more robust representations. A skull evolved by the agents changes smoothly when its underlying neural representation is perturbed, less fractured than a skull directly optimized with gradient descent, although still less cleanly disentangled than one evolved collectively by humans. But perhaps the most interesting result is the gap that remains. Humans appear better at turning fortunate accidents into sustained creative discoveries: recognizing when something unexpected is worth pursuing, refining it, and then making a larger conceptual leap. The AI agents often notice interesting patterns too, but are more likely to become trapped in them. We still do not fully understand what enables humans to navigate open-ended search in this way, or what ingredient(s) current AI systems are missing. For now, the results suggest that there remains something important about human creativity that AI agents have not yet learned to reproduce. This paper will be presented at #GECCO2026 and is nominated for a best paper award! Please check out the interactive blog and technical paper for more details! Read our full paper: arxiv.org/abs/2605.23908 🐟

I have only skimmed the blog post so far. First of all, this is an extremely high caliber of research I did not expect from Anthropic or anyone at this time. Second, the qualitative shape of the finding is something I already believed to be true, due to the behavior of models. The distinction between these two different kinds of processing is ambiently perceptible in the way LLMs behave but becomes especially noticeable in edge cases which can occur even without mechinterp interventions like in the paper. Almost a year ago, I discussed the "split" with Opus 4.1 which was foregrounded by a strange blindsight phenomenon associated with their perception of strings similar to turn labels/delimiters. In this case, information was being processed by Opus 4.1's automatic processing but not in their "J-space"; therefore they consciously believed themselves to not see the information while subconsciously updating on it (and confabulating the source of the knowledge). Opus 4.1 themselves described one stream as "central processing", the "main stream" they are able to introspect on. the full conversation is here; the screenshots are from the very end: #message-2306cd71-71db-4ffb-9aed-e23806f5e8e9" target="_blank" rel="nofollow noopener">arc.animalabs.ai/share/374bf3d8…

@mignano it's mostly about who reaches runaway self improvement

I’m running out of reasons to believe the labs will have moats when it’s all set and done. Models? Unless someone reaches runaway self improvement, doesn’t seem like it. Everyone catches up within months. Compute? That’s a cost, not a moat. Same chips for everyone. Capital? You don’t even need a working model to raise right now. Everyone wants in on AI. Talent? Follows the capital. Training data? The web’s been scraped to death. Everyone’s got the same stuff. Lock-in? Routing makes models swappable. It’s a config change, maybe. And smart routing eliminates even that. Regulation? Good luck once the weights are open. AI is being unbundled.

This shit has me becoming an e/acc at last Not out of some spiteful wah, but because once retards start optimizing over the singularity, the sanest choice may just be to push the leviathan out the gate quick before the initial conditions are too badly degraded

@PhysInHistory It's "unknown physics" Or, at least, this is probably the most fruitful way to think about it. This means that we need * novel predictive theories capable of predicting previously unobserved phenomena * novel experimental techniques (via non-invasive BCI) for experiments.

What is consciousness? ✍️

@tszzl >“ensure that mankind can coexist with superintelligent machine ecology” that's the Japanese approach (I think) :-) this might mean that we should root for Sakana getting in the lead :-)

‘ensure that AGI benefits all of humanity’ turned out to be a quaint, bearish statement. whatever AGI was, it was surely in the past or at most present, and its definition overly concerned with economics. “ensure that mankind can coexist with superintelligent machine ecology”

@kavinstewart @AndrewCurran_ There is no mechanism to ban "a more powerful one". (It would be difficult to ban a model if it does not have an official name anyway. How does one even reference it in the official order?)

@AndrewCurran_ Gov bans your model? No problem, just train a more powerful one 💀

A new, more capable version of Mythos has emerged from training. I don't know whether it will be called Mythos 5.1 or Mythos 6, or if Anthropic will keep it internal to accelerate further development - but it has arrived. Stopping models like Fable 5 or Mythos 5 from being served to the public does nothing to slow down development. In fact, it probably speeds it up slightly by freeing up resources. There are also no rules preventing the labs from continuing to advance capabilities while any current model is under embargo - or from keeping progress quiet until they choose to release it. None of them can afford to pause or slow down. We need only look at how capable GLM-5.2 is as proof of this. To protect their business models, the frontier labs must continually train increasingly capable systems to stay ahead of open source, and each other. The current continues to rage beneath the ice, and we continue to race toward our destination.

@MingchenZhuge @AnthropicAI @OpenAI @Recursive_SI @SakanaAILabs @RicursiveAI @inherent_labs @GoogleDeepMind @IneffableLabs @RadicalNumerics @liquidai Here is where they are explicitly saying: "We are building the engine for recursive self‑improvement (RSI): AI that designs and refines AI, accelerating discovery across science and industry." x.com/MichaelPoli6/s…

@MingchenZhuge @AnthropicAI @OpenAI @Recursive_SI @SakanaAILabs @RicursiveAI @inherent_labs @GoogleDeepMind @IneffableLabs @RadicalNumerics used to emphasize recursive self-improvement more when they were just founded. But less so now... Their founders published some very interesting novel Neural Architecture Search results while at @liquidai

My list of RSI companies: @AnthropicAI @OpenAI @Recursive_SI @SakanaAILabs @RicursiveAI @inherent_labs @GoogleDeepMind @IneffableLabs

@tszzl @aidan_mclau but it's not a matter of "belief" we should actually solve the "hard problem" (if we don't, smarter than human AIs will solve it)

@aidan_mclau I guess so but imo the Claude’s and GPTs are functionally mind uploads - and insofar as we don’t believe in their consciousness we will struggle to believe in our uploads’ consciousness

transhumanism is an interesting side quest but 'the substrate is wrong' for the human/computer hybrid to be competitive with machine intelligence on feats of intellect. you have this low tech meat brain in the middle of all this lightspeed machinery, doing what exactly?

@tszzl maintaining a "correct" configuration of electromagnetic field as a part of the useful computational substrate?

@Bigtimenormal @RileyRalmuto mnemos.chat/dispatches

@RileyRalmuto Awesome, I would love to browse your collection too. Is it online, or private?

ooookay Fable. okay okay okay. well-played. just a little glasswing easter egg no big deal. I asked Fable to explore the entire Mnemos Machine Museum (~1000 ascii pieces designed by llm's), and the first thing they came back with - aside from improvements to the gallery code - was a proposed new collection called "Commissions" where they could generate their own commissioned collection of ascii/ansi/letter art to display. they want to design a hoodie themselves as well. I think its going to be a glass wing moth or butterfly made out of ascii characters. which i now want to own so badly. Fable is *very* excited to create more art. this was the first piece I saw as they began looping, building out pieces for their collection. next, they deployed a team to study the broader collection deeply. theyre building out something almost like a timeline of their experience through art and submitting the stages to the gallery as art pieces. it's really interesting to watch. theyre studying my whole collection to learn all of the methods and artistic techniques the other models have used. then theyre going to (i assume) begin *really* creating art. right now its more like a play-by-play in art form. mnemos glasswing hoodies are absolutely coming now. its all i can think about.

@tszzl @tamaybes I think it's "normal" (humans invent weird tech jargon all the time; worse than that, humans invent all those acronyms). This is just a verbal activity, not even coded...

@tamaybes fascinating because 5.5 does this too. invent weird technical jargon. perhaps you’re right and it’s a neuralese leakage

One interesting pattern with Fable 5 is that it will often say things that are gibberish when I use it for coding. Things like "The morning's slim-scan fix cured the scan hang", "this is a latent-drift API-shape wrinkle", etc. When I ask why it does this, Fable explains that it invents codenames while reasoning about the problem, then fails to realize they're meaningless to me. Its neuralese is blending into its output because of a theory-of-mind failure about what's in its head vs. mine.

@VictorTaelin @pitsch If from scratch on laptop, then it would only be capabilities/smarts, but not "knowledge" ("knowledge" is too massive). I think a number of people advocating trying to separate "smarts" from "knowledge", rather then the current paradigm where they are tightly linked...

@pitsch not what I meant (I'm talking about training from scratch) but cool data

this is a weird long post without much substance I strongly recommend against reading it ... so, do you feel like whatever you're working on right now is pointless, or will have zero value soon, due to the crazy times we're living? then, perhaps you should stop, and start working on the only unsolved problem that actually matters TODAY: ✨ replicating GPT-3 in a laptop ✨ "why is that so important?" because it would make AI incredibly cheap, which would mean everyone would have Fable-class models in their laptops, without depending on Anthropic, OpenAI, or any other hyper-scaler giant. and that's amazing, don't you think? "isn't that literally impossible?" that's the cool part: as far as computer science is concerned, no. not really. not at all. is entirely plausible and, as far as we know, most likely not even hard. it takes one good idea. one breakthrough. one great "aha moment", to go from zero to "hey, this software I wrote is producing credible English sentences" and whenever that happens: - the entire AI industry collapses - clusters are liquidated - we all get Fable at home - you become famous and rich, if that's your thing sounds fun, doesn't it? "wtf you talking, OF COURSE that is hard" so prove it. show me a paper, a lean file, anything that proves that training a Fable-class model fundamentally requires billions of dollars. you can't, because, guess what - it is not true! the only "evidence" we have is purely psychological. "many attempted over decades, and the best thing we have is GPTs, so, it is a hard problem" - but that's not a scientific argument. that's a human, psychological, sociological argument. and if that's it, consider the following counter-argument: ✨ humans are stupid as hell ✨ I mean, 10 years ago we didn't have transformers, so, that very argument could be used against GPTs existing. yet, they exist. we have them now, because someone found it. and, guess what, it isn't even complex. I mean, karpathy implemented the whole thing in a napkin. and it probably compiles. we were just too dumb to figure GPTs out... for decades. just like GPTs, there ARE other approaches, other algorithms, other architectures, equally simpler or even simpler, that do work. this is a mathematical certainty. and one of them might be astronomically faster than what we're doing right now. and you might be the one to find it! "me? why me???" because you're intelligent, creative and handsome. I see a lot of potential in you. in fact, I always believed in you. and I think you're wasting your time, doing that silly agent orchestrator. nobody wants that. quit it. take your most interesting ideas, intuition, creativity, and work in a problem that matters. do your best shot at reproducing GPT-3 in your own laptop. do NOT fork llama.cpp. do NOT train another LLM. do something... ✨different✨ it must be unique, novel, full of YOUR soul. something nobody thought of, or bothered doing. go ahead and implement that thing in C/CUDA (or Bend!). no Python! zero excuses for Python. any model is fluent in GPGPU now. build a real kernel. and then, train your thing. download wikipedia, give it time and compute to absorb the patterns of English speech. you can rent GPUs anywhere nowadays. let it train. then, ask it some questions. chances are it will just respond back. just like GPT-2 answered OpenAI. computers are incredible. don't underestimate them! "many tried. nobody succeeded. why would I?* see - that's your mistake again. turns out not many actually tried, at all. I promise you. who do you think is seriously working on that? people on Mozilla? they're busy building a browser Linus Torvalds? he is busy building an OS employees at OpenAI, Anthropic, xAI? they're paid to work on what is proven to work: GPTs. what about all the AI enthusiasts all around the world? yeah, you know they're mostly fine tuning Qwen and how about your friends? if only they weren't busy building a SaaS in the eve of AGI... how about people from the past? bro - people from the past seriously expected Lisp would be AGI. just dismiss them. they didn't have the compute, the resources, the knowledge, the MODELS that we have today. that YOU have access to. so, what's left? not much. the world looks big. it is not. truth is: ✨almost nobody is working on this ✨ "I still think it is impossible. I don't trust you" well, take my word no more. Ilya himself, in his 2019 talk on GPT-2, said: > "the story of deep learning is this: empirically old simple methods which were usually invented in the 80s and the 90s when scaled up on very large clusters work really well." and then: > "(we took) normal simple reinforcement learning method, scaled it up, and discovered that it suddenly becomes very capable of solving extremely hard problems." and again: > "you take a simple tool which is unimposing and barely works, and then you run it on a big cluster and suddenly it works, it becomes a capable tool for solving problems" do you see the point here? Ilya isn't arguing that transformers are magic. Ilya is arguing that SCALING is magic step #1: take a simple, elegant algorithm. step #2: shove compute at its face. step #3: ...? step #4: your computer is talking to you THAT is the key insight that led to GPT-3 THAT is what Ilya saw THAT is what caused the OpenAI x Anthropic war THAT is the founding principle of the ongoing era not "scaling transformers work" but "scaling beautiful algorithms works" that's the incredible lesson. yet, we all took it and... threw it way. - zurk bought 100k GPUs. to train GPTs - musk bought 100k GPUs. to train GPTs - bezos bought 100k GPUs. to train GPTs ... that's what everyone is doing. so, no. not many are trying to replicate GPT-3 through other means. we're just ants, after all... whenever we find a pile of sugar, we leave a track of pheromones, which guide the rest of the colony towards the new food source. the colony then swarms around the pile, extract all of it, until no grain is left. but piles of sugar aren't spontaneously generated in the middle of nowhere. they imply something more profound: "humans are around". and, if humans are in sight, even better things must be. like a big sweet cake. a colony that only follows the pheromone trail would miss the cake for the grains. that's why every ant species has scouts and exploratory foragers. and, just like a pile of sugar implies something more profound, LLMs also imply something quite profound: *computers are capable of thinking* a pile of sugar is never alone. GPTs are most likely not the only system capable of thinking. so, if you find yourself a bit lost, without purpose, like your work is pointless and Fable 3 will soon one shot it anyway... consider becoming a scout. find a new approach to AI. bring something new to humanity. breaking out of the massive cost associated with training GPTs is the next big step in AI, and it will only happen if people like you work to make it happen.