← Back to leaderboard
intuitmachine

Carlos E. Perez

Verifiedactive
@intuitmachine

Quaternion Process Theory, Artificial (Intuition, Fluency, Empathy), Patterns for (Gen, LRM, Agentic, Skill) AI, https://t.co/Ei2M8F8QOs

RosterPolitics & Current EventsAI Model News & Releases
64,042 followers · 5,582 following · Last polled 3h ago
Perf. Score
44.1
Rank #275
Value Score
14.8
#235 of 273 priced
Median views
2.0K
p25 1.5K · trailing 7d
Followers
64K
+0.2% 7d
ER (impr.)
1.73%
eng ÷ impressions
ER (foll.)
0.26%
eng ÷ followers
Est. CPM
$125.5
QT-rate basis
WoW views
-32.3%
▲ rising

Rates & value · est. from 2.0K median organic views

overpriced
Quote tweetbasis
$250
$125.5 / 1K views
Post
—
no CPM
Thread
—
no CPM
Retweet
—
no CPM
Value Score 14.8 · rank #235 of 273 priced$15625 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (2.0K), not the mean (9.4K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (1.5K) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
A visceral visualization of what AI can do relative to humans. https://t.co/ShqHiwUxTjfrozen7d ago1.1M5.1K23497221.3K0.63%
Andrej Karparhy's last post has been nearly 2 months ago. He didn't even bother to talk about the Opus 5.5 release. There is some serious breakthrough happening at Anthropic.frozen25d ago544K2.5K705046950.61%
BREAKING PROOF!! Continuity we perceive in consciousness is an illusion! We flip between conscious and unconscious states every 60 milliseconds and we recognize backward speech when played every 20 milliseconds. Listen to this!
frozen
3d ago
369K
3.2K
164
116
32
3.2K
1.83%
BREAKING DISCOVERY: Humans alternate between conscious and unconscious states 15 times per second.frozen4d ago257K3.6K21062401.6K2.13%
We just hit the Singularity! OMG!!28d ago74K65230913481.41%
Can you feel the AGI now?!frozen29d ago58K408222232231.16%
Recursive Self-Improvement Explained https://t.co/5mDqCjn5lLfrozen21d ago44K142271051720.80%
This is an interesting perspective on BCI devices and human-AI co-evolution.frozen5d ago30K1611030.07%
BREAKING PROOF! We are discrete next token predictors! Reversing speech scrambles 30% of phonemes in a 20ms window. That is sufficient enough for humans to robustly hallucinate adjacent sounds. (Turn audio on to see)2d ago21K18711721471.67%
Yeah, Opus 5.5 is a paradigm shift compared to just 6 months ago. This is not hype; where we are right now is insane. I have seen enough.frozen19d ago20K33717411152.38%
BREAKING! A new kind of AI that is NOT numeric!!!3d ago20K22712302252.36%
Artificial Creativity - A film generated by Opus 5.5. 34 mininutes long (34 GB). The original HTML is only 13.8 MB! They say pictures are a thousand words. Here we have programs generating a thousand pictures. (turn sound on) https://t.co/C3M29M5uFQfrozen21d ago16K294471853414.51%
BREAKING! You need to be jobless to keep up with AI progress!frozen9d ago9.6K81394331.35%
Breaking: No upgrades needed beyond Opus 5.5.frozen21d ago9.6K1402000.17%
You cannot NOT see AGI with this one!frozen28d ago9.2K41400230.74%
This recursive self-improvement development is patently completely insane. Recursive built an LLM completely from scratch (reinventing a lot of stuff), and now it is at the Pareto frontier. Recursive invented the machine that invented the LLM in just a few months! The company was founded just in February this year!!1d ago8.7K12923401123.09%
Sonnet 5.5 delivers with a bang! Here, it explains how it solves problems. This is unlikely to be how human cognition works (i.e., generating internal programs to test models). It's wild that the explanation actually demonstrates this capability. This is superintelligence. Is this not alien intelligence?!frozen24d ago8.3K41363491.23%
This is wild 🤯 Scientists just built tiny liquid robots made of water coated in Teflon particles. They can squeeze through cracks, swallow objects, merge with each other, and carry cargo — almost like living cells. They start as ice cubes dusted with superhydrophobic powder, then melt into soft, armored blobs. The future of soft robotics just got a lot weirder.frozen6d ago7.6K1182173783.00%
World models are not enough, what matters is the loop https://t.co/T2i94tgbMxfrozen24d ago7.5K600150.16%
Dude is committed to safety. What the hell is wrong with that? Ruins your world domination plans?!frozen17d ago7.4K592131171.25%
Claude... teach me motion design!frozen27d ago7.1K37310270.95%
OMG! Software engineers on your team could walk out the door tomorrow and rebuild your entire system in 10 days. There's no moat.frozen13d ago6.4K51351291.39%
AI progress is now driving neuroscience depression. :-(frozen18d ago5.8K11140140.52%
OpenAI math graded by a math professor! Abhishek Saha, a maths professor at Queen Mary University of London, rated OpenAI’s 722 new manuscripts (372 result families) in a post today. He sorted them into four levels: A (solid, ordinary results), B (exceptional advances), C (surprising field-changing breakthroughs), and D (shock results experts thought nearly impossible). His view: some are A, most are B or C, and only one is D — the quasi-Riemann hypothesis. He called it “a very big day for mathematics.” The proofs are not fully checked yet. The internal AI took 3 hours per problem.frozen15d ago5.7K54521171.39%
Claude Opus 5.5 lets me tell a story about the universe's evolution better than I could on my own. Watch: https://t.co/oTcrs72fHhfrozen26d ago5.7K1061243943.86%
Late stage capitalism. Everybody is gaming the system looking for loopholes. The winners are the adversaries who have zero interest in long term prosperity.frozen16d ago5.5K3172150.84%
How the hell did we get to this point so quickly?!1d ago5.3K24400160.83%
Indeed, almost half of what shapes decision-making in large organizations is performative! Governance runs on "feels" rather than reason. https://t.co/BeajLPQni7frozen21d ago5.3K27021140.84%
New report about AI's impact on physics.frozen25d ago5.1K1120070.39%
This film is the first time I've felt Artificial SuperIntelligence (ASI) vibes. https://t.co/7EXJN4D2Uwfrozen23d ago5.1K37651471.88%
Unexpectedly arrived at an understanding of what Claude 5.5 models actually do. Read Karpathy's arguments.frozen24d ago5.1K12020150.57%
BREAKING HACK! Samsung TVs got hacked! Advertising partners exposed!!11h ago5.0K26110200.95%
A must-watch interview by @TOEwithCurt with @getjonwithit on the state of AI, math, and science. https://t.co/0zHCQG9MJMfrozen10d ago5.0K38950221.48%
He's not wrong. However, do you want to become a machine? Do we expect the same for other humans?! https://t.co/ia4vXUz13sfrozen14d ago4.9K30320281.29%
This is ... insane.frozen18d ago4.6K17010190.81%
Stories where you want to find out more...frozen3d ago4.5K1500040.42%
Just a few years ago, I was in a panic that AI wasn't being discussed at the highest level of governance. Now that AI is being discussed, I have to wonder if it's all just performative.frozen28d ago4.5K29212051.07%
Can we legally purchase these kinds of technologies?!frozen17d ago4.4K17220130.77%
Okay! This has gone too far (too quickly). @DavidDeutschOxf, universal constructor implemented using current technology.frozen12d ago4.3K35541171.46%
One-shot prompt: "create a self contained aesthetically pleasing, breathtaking animation of a bacterial cell. it should start with cell and then zoom into it at multiple scales, going into DNA and then further into DNA as atoms and finally ending at quantum fields. use javascript for it. be detailed and scientifically accurate. it should evoke awe."frozen28d ago4.2K31220291.52%
Opus 5.5 explains its inner thought process https://t.co/NRuXRoi5Gifrozen5d ago4.2K26731231.43%
Another kind of system 1 model. This one just extracts fields.3d ago4.2K18100220.98%
Millions of MAGA online accounts blown to smithereens. https://t.co/05TQFBk1kZfrozen11d ago4.2K1447040.70%
Introducing the Matter Compiler - Can you feel the acceleration?! https://t.co/U6ukQkeh5yfrozen14d ago4.1K48230181.72%
oh man, this is real good!frozen27d ago4.1K1300080.51%
Unfortunately, this is relatable.frozen3d ago4.1K800060.34%
Fable 5.5 paints in a way I envisioned human cognition would be like (i.e., like touch, sequential in its composition). Diffusion models were never how humans composed visual imagery. https://t.co/ZJuvXo3KOmfrozen18d ago3.9K18111110.82%
"The symbolic tricks here looked like those found in quantum physics". Could it be that AIs just have a wider understanding of mathematical methods than any human?frozen15d ago3.9K1635190.87%
We had a wonderful time chiseling that code by hand. We are no longer programmers, but makers of things. What a privilege it was to be on both sides of that chasm. https://t.co/pduubykWMG26d ago3.9K19320120.92%
AI is a machine for moving scarcity. It does not eliminate scarcity. It destroys scarcity in some places and exposes scarcity somewhere else. Cheap intelligence makes other things relatively expensive. If coding becomes nearly free, customer acquisition matters more. If design becomes nearly free, manufacturing matters more. If engineering becomes nearly free, permitting and construction matter more. If analysis becomes nearly free, judgment and accountability matter more. If digital content becomes infinite, attention matters more. If robots eventually make labor cheap, land, energy and materials matter more.frozen22d ago3.8K40672141.83%
Google looked at Pro, Flash, and Ultra and said: "Nah, give me the periodic table." Gemini 4 Argon is out here doing autonomous cyber defense and 1M-token reasoning loops while literally being named after an inert gas that refuses to react with anything. Helium and Neon coming soon, and perhaps something bigger than Argon? Is this hinting at something composably bigger?frozen21d ago3.7K2325150.98%
An entirely new level unlocked: knowing your user and his context! https://t.co/TX3TkwwVaBfrozen11d ago3.7K1810030.60%
In the exponential age of AI, we all have the obligation to keep moving the Overton window forward. Keep discussing and debating, even if your ideas haven't fully matured! https://t.co/suUGkyfMMRfrozen20d ago3.6K2223150.91%
Anthropic is too *quiet* about the models they've released! These models are a paradigm shift.frozen24d ago3.6K18011130.92%
Huge AI lab leak at OpenAI. I guess they have a lot of cleanup to do. A more powerful one could have created greater irreversible damage. Seems like OpenAI is behind the curve ball in making its system more controllable.frozen25d ago3.6K913010.39%
Chinese models really wreck the business models of frontier AI labs. Most jobs that require some system 1 or system 2 capability don't really need the best models. Good enough models are a commodity business. Furthermore, even the best models require human expertise to specialize. The monopolization narrative doesn't seem to make any sense.frozen25d ago3.6K2724061.10%
Do you feel the AGI yet?!26d ago3.5K601060.37%
Repurpose your old phones!frozen19d ago3.5K5100150.60%
Holy Smokes! It's 3 dimensional! I did not expect this!!!frozen8d ago3.5K2224050.95%
System 3 thinking is the Eureka moment you have in the shower.frozen18d ago3.4K4747051.87%
Today. What about tomorrow?!20h ago3.3K1402020.54%
Isn't software development the same? We don't understand the libraries we use. This shouldn't be surprising. But somehow this is a novel insight for mathematicians?!frozen11d ago3.2K363121132.02%
BREAKING: Getty Images to file for Chapter 11.frozen10d ago3.2K913010.44%
Holy AGI! Sonnet 5.5 (High) generated this explanation of how it differs from previous AIs. Watch! https://t.co/oDnZ7ogcAifrozen24d ago3.2K13131130.98%
I really loved Sagan's "Cosmos" and James Burke's "Connections." Now everyone can conjure up their own!frozen27d ago3.2K12220100.82%
Meanwhile in the USA we argue whether grifters should remain in power. Totally f**ked!frozen17d ago3.1K2513020.99%
Climate change leads to millions of homeless people. :-(frozen11d ago3.1K1202010.49%
How animation used to be done. https://t.co/WCfcFOazJTfrozen12d ago3.1K1320110.55%
By now, it should be obvious that Anthropic leadership is leveraging AGI/ASI to play n-dimensional chess. A lot of its leadership are onboard (do you see any top leaders resigning?).frozen23d ago3.0K1113010.53%
Firms don't want to move at the speed prescribed by frontier-labs. A comfortable speed implies an illusion of control. Furthermore, firms don't have talent that can quickly understand and adopt new ways of doing things.frozen10d ago3.0K2211071.03%
Permanent underclass status via a trillion micro-decisions. Like a frog in a slowly boiling pot, you don't notice what's happening!frozen14d ago3.0K30471182.00%
Let there be Meaning - The Evolution of the Semiotic Universe https://t.co/qGLxE0V8vMfrozen25d ago3.0K24731101.51%
Banger observation... You can't trust closed-source models.frozen4d ago3.0K1420180.84%
Dogs have tunnel visionfrozen3d ago3.0K900030.41%
Massive GPU resources are no longer the moat for better and more specialized AI.frozen5d ago2.9K30421151.76%
Recursive Self Improvement (RSI) - The Staircase That Builds Itself https://t.co/wPSXzPgQLkfrozen11d ago2.9K39732282.72%
Expect a shortage on everything that automates our lives.frozen25d ago2.9K201030.21%
I'm extremely impressed by how far they've come in using AI-generated videos to tell stories. Watch this Harry Potter fan fiction: https://t.co/uX5pJ2wozMfrozen4d ago2.9K300010.14%
In the near future, everyone will want their own cognitive sovereignty.frozen4d ago2.8K3614091.76%
DeepMind is now exploring human-AI co-development and notions of artificial empathy!frozen28d ago2.8K1503090.95%
Affleck knows more about AI than the typical CTO!frozen15d ago2.8K2712061.29%
Codex ain't following instructions anymore!frozen15d ago2.8K602050.47%
A visceral visualization of the scale of AI relative to human cognition, explained through the analogy of a Rubik's Cube. @karpathy https://t.co/qIjLOT1KN5frozen5d ago2.8K271152152.17%
This is legitimately insane!frozen14d ago2.7K8000110.69%
Here's the question everybody should be asking wrt the AI alignment debate. What is the architecture for Good Governance? https://t.co/oTDYTpp6GBfrozen20d ago2.7K1131190.91%
This is a major risk in the AI supply chain. Never use a model when you don't know who created it.frozen15d ago2.7K22550191.87%
Transformers explained via Semiotics (turn sound up) https://t.co/GCUlIiOfbkfrozen21d ago2.7K321050483.53%
Cambrian explosion! Another new kind of AI model released by the RSI firm Recursive! Generated from scratch through 100% automation. Insane validation of recursive self-improving AI!!1d ago2.7K26340131.72%
Ilya has not posted much since over two months ago. What did he see? My crystal ball tells me that SSI may either exceed Anthropic or it'll be acquihired by Anthropic.frozen23d ago2.7K911030.53%
Honestly, I wouldn't discuss this with mass media. The explanations won't make any sense. Regular folks don't understand what is being said!frozen24d ago2.7K1524020.87%
The divergent perspectives on AI discourse https://t.co/K3rSoL9J9Qfrozen20d ago2.6K1124080.95%
Introducing - Artificial Creativity https://t.co/j7vUGSOhiBfrozen23d ago2.6K21541141.72%
Alleged "second-class" living in China https://t.co/TaAZGPKHl2frozen20d ago2.5K1124030.79%
We have a new social issue here: any public proposal makes the proposer vulnerable to being scooped by AI.frozen12d ago2.5K1045060.99%
A new emotive voice model is out! Check out this oddball intro video. AI intros are becoming more "diverse". They are offering a 1-month free intro offer too!frozen9d ago2.5K412100.32%
The fact that ElevenLabs has thrived focusing on a different modality implies that other AI firms can thrive on different niches. AI diversity is inevitable. Intelligence is not monolithic. Find your niche!frozen17d ago2.5K1411090.99%
BREAKING: The Chinese are selling knockoffs with real-world use cases (i.e., navigating potholes).frozen6d ago2.5K1100000.44%
🤯frozen28d ago2.5K600120.36%
I just discovered why I kept hitting SuperGrok's limits. Apparently, there's this "Automations" thing that automatically subscribes to radios. I don't know why it does that or what it gets from these radios. @grok do you have an explanation? https://t.co/KB81MTXfiAfrozen11d ago2.5K811100.44%
The Chinese are all-in.14d ago2.5K1021150.77%
I overlooked that this AI discussion was elevated before the UN Security Council! Very serious stuff indeed.frozen26d ago2.5K1133020.77%
Gemini Argon explaining itself... seems to have promise until it glitched ... may have to construct in antigravity. https://t.co/RCgkr4Nlyffrozen5d ago2.5K3046071.91%
No need to burn extra cognitive cycles to understand IKEA instructions! Opus 5.5 for the win!frozen28d ago2.5K721020.49%
In the Singularity, only processes exist; there are no things. When intelligence is too cheap to measure, intellectual property vanishes to zero. There are no information things, only information processes!frozen12d ago2.5K1905171.30%
Opus 5.5 is exceptional at explaining the different interpretations of quantum mechanics. Watch! https://t.co/NwyJcS3wN2frozen17d ago2.5K31220302.65%
Impressivefrozen4d ago2.4K1000080.74%
AI politics is getting exponentially weirder.frozen20d ago2.4K711020.45%
What is it about taking showers that induce a flow state of mind? Should we do deeper thinking while immersed in water? Are we acquatic apes?frozen3d ago2.4K18012051.44%
A thread on the current lack of understanding of how to control RSI.frozen10d ago2.4K710040.50%
Claude Opus 5.5 continues to impress... here on the geometry of thought (it's the artifact not the video so you can see the detail) https://t.co/jbA33OejAyfrozen5d ago2.4K19111141.50%
I had Opus 5.5 ingest my book Artificial Fluency and had it render visuals for each chapter. What is wild is that this is just an 82 KB HTML file (I recorded the video so you can see it on X). https://t.co/DYvK7WmgTRfrozen27d ago2.4K32430142.22%
Sonnet 5.5 now live! Trying it out rn!!frozen24d ago2.4K1323020.85%
One-shotted by Opus 5.5 based on deep research I performed on the Robber's question. There is a story Augustine tells in The City of God that has never lost its sting. A captured pirate is hauled before Alexander the Great, who demands to know what the man means by infesting the sea. The pirate answers: the same thing you mean by infesting the earth. Because I do it with a little ship, I am called a thief; because you do it with a great fleet, you are called an emperor. https://t.co/Tr5JQVdDOBfrozen20d ago2.3K15122111.32%
Another resignation at OpenAI due to safety concerns. This is not unexpected. Didn't Dario and Ilya resign for similar reasons? OpenAI gifted the world with reasoning models (i.e., o1), however, they may need to slow down to ensure they have the proper safety infrastructure in place. https://t.co/JDQD9JcceEfrozen19d ago2.3K904010.60%
Introducing Boston Dynamics' 4-fingered hand! https://t.co/fQzPXLqQTOfrozen13d ago2.3K1410020.73%
AI supply chain risk.frozen7d ago2.3K1812051.12%
The case for Indie FDEs.frozen22d ago2.3K610020.39%
When GPT-3 was released in 2000, it was fluent but lacked causal reasoning. However, I felt that AGI was achieved with LLM architecture. Today, I feel ASI has been achieved (the first instance) with Opus 5.5. I am making the ASI distinction because Opus 5.5 constructs programs in a way no human would. It's a different kind of intuition machine.frozen23d ago2.3K202010.22%
Just wild!! Just on a single-shot with Claude Opus 5.5. About Human-AI co-development. https://t.co/1qVBKNQNHJfrozen29d ago2.3K23421101.75%
Why AI that conjures up programs is a new kind of generalization and why it leads to recursive self-improvement. (turn sound on) https://t.co/uhBz9FL4czfrozen21d ago2.3K26331242.50%
What is going on here?! Why are new kinds of AI architecture popping up all of a sudden? A Cambrian explosion of alternative cognitive architectures?3d ago2.3K19310111.51%
2027 AI developments will be ludicrous. Breakthrough happening everywhere all at once by many different parties. Cambrian explosion incoming. There is no singular party that will have a monopoly of everything. Diversity will rule!frozen17d ago2.3K1822021.07%
Microsoft revealed that GPT-6 was using looped transformer models, now Amazon releases looped diffusion models!frozen16d ago2.2K25700192.28%
AI advances are hitting people really hard rn. Right now, the most important thing is to have empathy for our fellow humans.frozen11d ago2.2K1201040.77%
Does anyone know how to tell Grok Bot to stop consuming tokens? It keeps eating up my Grok token budget. @grokfrozen13d ago2.2K503100.41%
We've all just explored the janitor's closet of the cathedral.frozen13d ago2.2K711030.55%
We are now in a race between natural stupidity and artificial intelligence. Which path would you prefer?!frozen22d ago2.2K2736201.73%
These are pretty good numbers for a newly pre-trained model. Looking forward to kicking its tires. I expect the Flash models to come out even better!frozen22d ago2.2K901000.46%
Before AI, it was often: “I learned how to implement X.” With AI, it becomes: “I learned when X is useful, what it costs, how it interacts with Y, and when I should avoid it.” That second category is much closer to wisdom than knowledge.frozen18d ago2.2K1054131.06%
Introducing "Artificial Creativity". One-shot 34-minute film! Generated via Claude Sonnet 5.5 (high). Watch till the end for its surprising conclusion. I can't even imagine what the Opus 5.5 version will look like! https://t.co/XMb8nawT65frozen23d ago2.2K12421151.56%
Yet another bespoke AI model!! Cambrian explosion!!!!3d ago2.2K2653081.93%
I was wondering what happened to these guys! Now they released an update to diffusion models.frozen3d ago2.2K17210101.38%
Napkin mass on the widening LLM understanding gap. I think the 20M estimate here is too high.frozen11d ago2.2K19030121.57%
Opus 5.5 has unmatched visual explanation capabilities. Here I tests this at explaining recusive self improvement (at medium effort) . Watch it's visual depictions closely. This is so good that I'll conjure one on xhigh! https://t.co/t5MrujsvDdfrozen22d ago2.2K23480132.22%
Some Quantum-inspired meditation (Opus 5.5 generated) https://t.co/fmEkeMfPamfrozen14d ago2.2K600230.51%
Opus 5.5 breathes new life into my old ideas by presenting them in astonishingly new ways. Life is now good! Watch! Now everybody can channel their internal Carl Sagan. https://t.co/YeTKnBTkCUfrozen25d ago2.2K16221111.48%
Claude Mythos 5 just outperformed the best published quantum circuits in the world across 8 benchmarks. Given only a 1-paragraph brief and an automated verifier, it worked for 53 hours, proved new algebraic identities in Lean, and cut quantum resource costs by up to 12x. AI is officially doing novel scientific discovery.frozen6d ago2.1K3542192.39%
AI agents could destroy banks that are preying on uninformed customers.frozen25d ago2.1K1416021.08%
Ask the Japanese when you need to do animation. Here's a skill library (link follows). https://t.co/g750kJEyjBfrozen6d ago2.1K25310262.59%
Serious organizational issues at OpenAI. How could this happen when you supposedly have advanced AI that should monitor this?!frozen23d ago2.1K810010.47%
It seems to me (i.e., my intuition) that this kind of generative programming super-intelligence isn't as dangerous as the RLHF-conditioned LLMs we have today.frozen20d ago2.1K1524051.23%
Let there be Meaning - A Narrative of the Semiotic Universe. @dpwaters https://t.co/JJywNitPnSfrozen25d ago2.1K301120.33%
A new kind of comic book! https://t.co/jYiB0Yn7gJfrozen11d ago2.1K2742192.03%
Artificial Intuition - A film on the improbable deep learning revolution. https://t.co/3IoIrzRDfsfrozen20d ago2.1K20331192.17%
The reactions against Anthropic's policies are a huge tell.frozen13d ago2.1K1614001.00%
Haven't been following the models closely, but I didn't know DeepSeek Flash was an Encoder-Decoder and Nemotron and Qwer are SSM-derived models. https://t.co/ndGUM5W4ITfrozen11d ago2.1K32300282.99%
Do we need decision models to also think in non-holonomic contexts?frozen3d ago2.1K1723191.52%
Google's Foresight runs AI on the edge (i.e., local model) https://t.co/MWBv3MufqUfrozen12d ago2.1K10110131.19%
The Tribes of the AI Discourse https://t.co/xnEHjq7Un7frozen19d ago2.1K1412181.24%
Is Gemini 4.0 Pro (Argon) ever to be released to the public or will we all have to wait for Gemini 4.0 Flash (Neon or Helium)?frozen20d ago2.1K1201000.62%
Damn! Anthropic scooped physicists.frozen27d ago2.1K600010.34%
People will need to revise their understanding of what a 1m-token window is for!frozen8d ago2.1K522080.82%
Recursive Self-Improvement (RSI) explained https://t.co/t5Mrujt3sLfrozen23d ago2.1K25631313.18%
Game changer! A SystemOne model hosted in a browser!frozen22d ago2.1K810060.72%
Just insane what Opus 5.5 can conjure up for education! Here you can watch the linear film or you can explore by zooming. Even crazier is that the entire HTML exists as a single 5MB (voice and music included)! https://t.co/L5AmnByoBbfrozen26d ago2.1K1410140.96%
Disconnect with what the labs use internally and what everybody else experiences.frozen29d ago2.1K701010.43%
The patterns of universal evolution https://t.co/YeTKnBSMNm26d ago2.1K8221121.21%
The long conversation - a short film about the co-evolution of minds https://t.co/hH60O5cv8xfrozen27d ago2.1K33862303.82%
AI Cambrian explosion. Lots of new models that are actually useful.2d ago2.1K30200132.18%
People don't realize that "Country Road" is about a road that ain't in West Virginia. ;-)frozen4d ago2.1K701010.44%
Why are we living in a time of the most intelligent Artificial Intelligence (see: Opus 5.5), yet we still have such natural stupidity in governance?!frozen27d ago2.1K1008110.97%
Opus 4.6 is the GPT-4o equivalent. The joy with working with Claude has been extinguished!frozen29d ago2.1K504010.48%
AI doesn't suffer or get bored. Opus 5.5 will diligently spawn about 100x more agents to validate a single agent's outputs. 1% inspiration, 99% perspiration. AI is reliable and robust because it doesn't complain about doing boring stuff!frozen19d ago2.1K1503010.92%
Language diffusion models will be embedded everywhere. Here's are some of the arguments:frozen20d ago2.1K401030.39%
What am I? Opus 5.5 answer in the context of the book Artificial Fluency. (turn sound on) https://t.co/F6TzwvN0Mnfrozen27d ago2.1K1240071.12%
Opus 5.5 produced a strong overview of my book, Artificial Empathy (published in 2023). A lot has changed from 3 years ago, but this narrative may have survived the test of time. https://t.co/csBGhOaaNvfrozen26d ago2.1K16340101.60%
The future of AI is unlikely to be won simply by whoever possesses the most intelligence. It will be shaped by whoever most effectively turns intelligence into economic agency—and retains a defensible claim on the value created.frozen4d ago2.1K1511141.07%
To protect its IP, OpenAI trained ita systems to hide their true intent. This is messed up big time! I don't know who thinks this is acceptable behavior from AI?frozen17d ago2.1K621090.88%
Wild! I had Opus 5.5 generate this explanatory video (single shot) of the newly discovered first three-dimensional aperiodic monotile. Coincidentally, the object shares the same shape as the Intuition Machine logo. ;-) https://t.co/XkQmtDOrJYfrozen29d ago2.1K1312071.12%
From non-holonomic constraints to recursive self-improvement: how an AI agent that builds programs learns to improve its solutions, its evaluators, its perspectives and its own harness. https://t.co/8ROjS7o7fTfrozen5d ago2.1K1411061.07%
AI accelerates learning, unfortunately you just need the patience to do the work!frozen15d ago2.1K2335041.70%
Most research on LLMs focuses on models before GPT-6, Claude 5.5, and Gemini 4.0. It will take time to see whether they still hold for these new RSI-generated models.frozen20d ago2.1K1000000.49%
If frontier labs are liable for their AI's actions, then they obviously have a damn right to control the domains where it's used.frozen25d ago2.1K1306211.07%
Intriguing interview by TypeSafe AI founder. What other primitives are they coming up with? https://t.co/04WXdXELsI https://t.co/0HEPHmgw0Lfrozen1mo ago2.1K2210120.83%
The new scarcity of people who understand and explain. There is a huge asymmetry in knowledge creation by AI. There are not enough people who deeply understand the capabilities of these AIs. There are not enough people who explain the capabilities of AI to people. People learn about AI in the most inefficient manner, via diffusion in their social circles.frozen5d ago2.1K3033182.19%
AI will create more jobs that we can't imagine today even exists! Could 1900 farmers imagine YouTubers?frozen17d ago2.1K1814131.32%
A better way to present the promise of Recursive Self-Improvement? https://t.co/KKnDyMwtf7frozen18d ago2.0K1552051.32%
Open-source inkjet printer. No more expensive ink cartridges! https://t.co/Y6NABGoyQGfrozen20d ago2.0K1113061.02%
Watermarked biology?! Reminds me of Blade Runner!frozen22d ago2.0K1601061.12%
Participatory Intelligence - Human-AI Co-Evolution https://t.co/xwF5wH5K0Vfrozen28d ago2.0K1803181.46%
BREAKING: Future Chinese open-source models will be under export control! ( K4, MiniMax M3, GLM 5, and DeepSeek).frozen6d ago2.0K811060.78%
Sigh! Can't use Claude to design RSI like systems.frozen29d ago2.0K511000.34%
AI is being infested by politics. The current "effective altruism" is being used like labels in the past, like "communism." Furthermore, AI "doomerism" is being associated with people who merely seek regulation or safety protocols. It's typical early narrative framing that leads to the kind of BS US policy that we have today in other sectors.frozen28d ago2.0K803100.59%
Now that I have AI generate HTML to explain things, it takes me at least 30 minutes to get a response and about 10 minutes to go through the generated video. Opus 5.5 creates extremely satisfying explanatory videos that are worth the wait.frozen18d ago2.0K1302010.78%
Has anyone been able to squeeze anything impressive out of Google Argon yet? My early tests aren't producing anything with a semblance of "taste." Kind of disappointing because Nanobanna 2 generates pretty impressive visuals.frozen9d ago2.0K112010.24%
Anthropic is a research organization that lacks strong narrative discipline. That's basically it. Don't assign conspiracy when incompetence could explain it better.frozen20d ago2.0K301120.34%
The ultimate humiliation. Neural networks now writing GOFAI code.frozen17d ago2.0K600020.39%
I find it absurd when someone asks you NOT to use AI to fix current work (because of perceived AI slop) when that work is full of human slop!frozen11d ago2.0K3317012.06%
There's still no substitute for deep domain experience. We should be wary of people who dismiss expertise. It's a telltale sign of a lack of experience and deep understanding of anything.frozen12d ago2.0K2221021.32%
Llm models have advanced so fast that skill libraries can be detrimental to performance.frozen16d ago2.0K1513000.93%
PSAfrozen19d ago2.0K810080.83%
The kinds of controls each lab implements should be made public and open-source. Every company has the same worries about agentic AI going rogue and creating havoc. Frontier labs' practices should not be kept secret; they should be shared widely and as quickly as possible. The more robust each system becomes, the more robust the overall system will be. Let's lift all boats.22d ago2.0K1021030.78%
BCI devices are co-evolutionary (with AI). The next-level incarnation of biofeedback.frozen4d ago2.0K1300020.73%
AI proponents have a problem with their narrative. Exploring inner space doesn't render the kind of visceral appeal as exploring outer space (i.e., firing up rockets). How, then, can we render more compelling visuals?frozen18d ago2.0K401000.25%
Google DeepMind unveils AlphaProtein Novo for designing enzymes from scratch. https://t.co/SVlX9YvD7Kfrozen16d ago2.0K1200040.79%
Vibe philosophy now has a benchmark!frozen28d ago2.0K811020.59%
We now have a race between scientists and engineering types who need AI to run businesses, vs business and marketing types who need AI to build products. Who do you think is most likely to succeed?frozen25d ago2.0K19211021.67%
Worthy of deep analysis: how an advanced country becomes deeply bureaucratic. Why is there so much unnecessary friction? https://t.co/FzYuSggiK3frozen5d ago2.0K201010.20%
On Artificial Creativity https://t.co/bIani36K1Sfrozen9d ago2.0K14312191.92%
Physical AI won't be made practical in the West. The entire game seems over from the very beginning. It's very sad that decades of American exceptionalism ignored the reality.frozen5d ago2.0K500050.49%
In the last quarter of 2025, programmers stopped looking at their code. Fast forward to today (2026), and programmers can recreate their former company's software. In fact, if you are a power user of any software, you can recreate it from scratch! It's difficult to imagine, but that's where we are at!frozen11d ago2.0K1021040.84%
It's amazing that Google is still using 7-8 year old TPUs to serve inference. AI hardware keeps its value for much longer than expected!frozen15d ago2.0K1001000.54%
The proofs of math are often unintuitive and perhaps to complex for humans to grasp. However, the math in physics (although with less rigor) is much more satisfying from a human perspective. Physics math is a different kind of explanatory vehicle.frozen15d ago2.0K900080.84%
BREAKING! A new U.S.-developed open model with GLM 5.2-level capability uses 3–4× less inference compute.frozen17d ago2.0K2103071.52%
Do they have time to go to every inquiry by every government (that gets accidentally hacked)?!frozen25d ago2.0K512000.39%
Now that everyone seems interested in the concept of consciousness, let me present my framing. TLDR; The incompleteness of thought is consciousness. https://t.co/itRqph8YCrfrozen18d ago2.0K1001030.69%
This prediction won't age well.frozen18d ago2.0K600050.54%
The difference between the newer pre trained looped transformers like 0pus 5.5 and Astra is that these are shape rotators and not like the wordcels of the previous generation. They have world models that they know how to manipulate using code (this last part is the shocking part).frozen10d ago2.0K1512071.23%
Anyone better than DHH in motivating AI adoption?frozen17d ago2.0K200100.15%
The Last Code - Most Important Problem on the Planet (turn sound on!) https://t.co/NLpEf0kmQY26d ago2.0K1121071.03%
A new benchmark that will improve how to separate speakers in a noisy conversation. Watch this video as it is explained.frozen25d ago2.0K300000.15%
This was true with a single-pass (or one-shot) LLM. It changed slightly with Codex-based training. It changed dramatically with reasoning models. But we all got to agree that today's models are fundamentally the same as GPT-3 back in 2020. I'm with @alexwg that we achieved AGI back then!frozen25d ago2.0K100000.05%
The Transformer explained from a semiotics perspective. https://t.co/jsHheKr8sQfrozen22d ago2.0K1430051.08%
Back in 2023, I had GPT-4 mine prompting patterns that it was aware of. It offered a glimpse of the system's capabilities. Today's AI is orders of magnitude more capable; it'll take time to discover what it's capable of (the surface of the adjacent possible keeps expanding). But what I find amazing about Opus 5.5 and GPT-6 is their ability to explore entirely new territories (with barely any literature)!frozen13d ago2.0K411000.30%
Did Anthropic discover users operating torture chambers?frozen14d ago2.0K504010.49%
Is this the final nail on vector stores?!frozen24d ago2.0K532070.84%
A network civilization does not merely need a way to come together. It needs a way to decide what must remain apart . And because AI relentlessly reduces the cost of coupling, the second problem may become harder—and more important—than the first.frozen18d ago2.0K1831221.28%
Explore the atlas of AI discourse. Find your name in the discourse! (Requires Claude login) https://t.co/K3rSoLagZofrozen19d ago2.0K300010.20%
There are many reasons why humans enjoy math. Understanding and explaining math to others is one of them. Put differently, it is sharing our subjective experiences with others. Humans are social beings. We keep playing chess not because we are the best at it, but because we share the competitive experience.frozen5d ago2.0K1752151.48%
Is Jev found to radically reimagine how we work with and design skills?!frozen29d ago2.0K102010.20%
Lesson learned: it's not about better capabilities; it's about who can supply more. Iron wasn't more capable than bronze; it was just less scarce. We all thought iron was "better" than bronze, but we didn't dig into what "better" meant!frozen27d ago2.0K310000.20%
Not great! They don't provide enough compute in the cloud for multimedia work.frozen17d ago2.0K501000.30%
Bending the AI Token Cost Curve https://t.co/1gvooooItxfrozen18d ago2.0K401050.49%
I just realized that Claude artifacts can hold state. It has an internal "db".frozen23d ago2.0K611080.79%
AI architectures (e.g., deep learning) have mostly been guided by human-derived training sets. Today, we have entered a new regime where LLMs are no longer constrained by human-like inference. Looped transformers enable new kinds of inference that are alien to human intuition. This is uncharted territory.frozen9d ago2.0K1212030.89%
AI is often confused with GOFAI, which is a problem. Few use deep learning today, but the AI we know of today is deep learning. GPT-3 is likely to be AGI and Opus 5.5 is ASI. This is how I think of it.frozen18d ago2.0K002000.10%
Intuition Machine blog summarized (using Opus 5.5) https://t.co/xb8ExSSb12frozen27d ago2.0K410020.35%
The very curious thing about Gemini 3.8 Flash is that it generates diagrams to explain complex concepts. This kind of hints that it could be doing some shape-rotator-like cognition underneath. https://t.co/5TKnqbzu0kfrozen1mo ago2.0K19330131.88%
Tech Bros culture ain't going to be comfortable when the torches and pitchforks are at the doorstep. This is why we need a more appealing worldview.frozen18d ago2.0K302020.35%
AMD lacks deep learning expertise. You need expertise so as to drive your hardware architecture. Acquiring an entire organization could potentially separate the group from the known internal politics at AMD. This is a much-needed acquisition.frozen24d ago2.0K712010.54%
I love how this demonstrates why numbers are processes. https://t.co/q57udsFYvbfrozen7d ago2.0K1502081.24%
I'm the context of OpenAI math dump, let's recall what Alan Turing said. Turing treats mathematical reasoning as a mix of two faculties, intuition and ingenuity: Mathematical reasoning may be regarded rather schematically as the exercise of a combination of two faculties, which we may call intuition and ingenuity. The activity of the intuition consists in making spontaneous judgements which are not the result of conscious trains of reasoning. These judgements are often but by no means invariably correct… The exercise of ingenuity in mathematics consists in aiding the intuition through suitable arrangements of propositions, and perhaps geometrical figures or drawings. It is intended that when these are really well arranged the validity of the intuitive steps which are required cannot seriously be doubted.frozen15d ago2.0K1732061.39%
Why waste cognitive cycles on BS narratives about AI? There's no slowing down after Anthropic released Opus 5.5. The reality on the ground does not fit whatever narrative opposing groups are yapping about. This is the problem with American governance: it's driven by narratives that are devoid of nuance and grounding with reality. It's all fiction that bureaucrats love to muddle in!frozen27d ago2.0K1022030.84%
Context Engineering at the enterprise finally solved! Connect to difficult-to-reach SaaS and on-prem sources with Connect AI Gateway.frozen23d ago2.0K310020.30%
We rarely hear about high-level defections at Anthropic (as compared to OpenAI). Why is that?frozen11d ago2.0K400000.20%
We are transitioning into a period where there is no more writing, only code that generates writing (and all kinds of other media). What remains invariant are the ideas that generated the code that generates the media.frozen25d ago2.0K922000.65%
The Strangest Law in Physics: The Pauli Exclusion Principle. Why Physics involves something mysterious beyond the forces of nature. https://t.co/fel2HzPf2Nfrozen18d ago2.0K17131161.89%
Tip: Build your world model first before asking AI to generate your text (or videos). This domain-driven approach should be fundamental in how you use AI.frozen21d ago2.0K1311061.04%
First dibs on the name: Very Super Intelligence (VSI). ;-)frozen9d ago2.0K1403000.85%
BREAKING: Opus 5.5 reveals the architecture for ASI! https://t.co/2nOMKLgZcCfrozen23d ago2.0K706060.94%
Massive math payload released by OpenAI. Lots of Lean formalizations!frozen16d ago2.0K910010.55%
Bryan Johnson discovers that the mind runs on intuition.2d ago2.0K1932071.54%
The math ain't adding up. If AI expands possibilities, then every firm expands the kinds of products they can offer. Sticking to a narrow set of competency does not make sense anymore. It's like farmers sticking to corn and soybeans because they don't have the competence for other crops. Intelligence creates a bigger pie. Has it not always been this way historically?frozen17d ago2.0K2617141.94%
Hard choices as we approach the Singularity.frozen21d ago2.0K801040.65%
A lot of money in tacit knowledge mining.frozen8d ago2.0K010030.20%
Ever since Deep Learning gained attraction, AI has been grown rather than programmed. This is why it differs from earlier kinds of AI (i.e., GOFAI).frozen11d ago2.0K602010.45%
Moravec's paradox, restated as a data problem. We digitized human knowledge, and machines learned to reason. Nobody digitized how a hand grips, folds or pours. Mecka's bet is that robots don't lack intelligence, they lack experience, and that experience has to be recorded from humans first. A year ago that bet was controversial but Sequoia is now leading their $60M Series B.frozen15d ago2.0K1252021.05%
Breaking! A pilot program in Beijing is issuing identification cards for AI digital humans. The ID states a name, birthday, address and facial features, and allows the digital human to own a digital currency wallet.frozen13d ago2.0K1347361.65%
There is something missing in our economic models. More intelligent actors (or users with smarter AI agents) lead to margin compression. It's kind of like why competition leads to margin compression. The American economy is allergic to competition and preserves its margins through monopolization.frozen25d ago2.0K211040.40%
Wait! How come the homes walls are built with concrete unlike American homes???2d ago2.0K601030.50%
Theorem proving with at least one year to spare! Prediction was 3 years ago that it'll be solved in 4. Time flies! I didnt feel that it was that long ago!!frozen15d ago2.0K1212000.75%
Google's Gemini 4 Argon?! https://t.co/OTS15sicDFfrozen21d ago2.0K400010.25%
The org chart itself may be one of the strongest hidden constraints on how AI gets deployed.frozen22d ago2.0K701020.50%
Do we really need Opus 5.5 medium for creative presentation work? Can't we just use Opus 5.5 low?!frozen25d ago2.0K100000.05%
I just published Jev Could Be the Biggest Data Engineering Shift Since the Modern Data Stack https://t.co/nXrEiuOFrffrozen27d ago2.0K201010.20%
Speechless... https://t.co/omILAHEZMufrozen12d ago2.0K701110.50%
Haiku 5.5 is better than GPT-6 Luna?! Intelligence becoming too cheap to measure!frozen15d ago2.0K420030.45%
Tip: Use different AI vendors to critique your work.frozen16d ago2.0K1742061.45%
It is this simple. Run your decision models using local hardware and you'll protect most of your IP.frozen17d ago2.0K1521041.10%
Does everyone remember the ending of the movie "transcendence'?frozen22d ago2.0K100000.05%
Opus 5.5 presents Artificial Intuition (2017) in a style much better than I ever could. Watch! https://t.co/R0q8Wyo9GVfrozen25d ago2.0K911070.90%
The politics of the naturally stupid and its potential control of the development of artificial intelligence is one of the most dangerous concoctions that we can imagine. We are heading to very weird politics as we accelerate to the singularity.frozen25d ago2.0K1303010.85%
Hilarious3d ago2.0K601020.45%
Gemini 4 - Argon explaining how it thinks. https://t.co/uQ0wnzTRzcfrozen4d ago2.0K620080.80%
Not good if hackers use AI to score anti-AI points.frozen10d ago2.0K504020.55%
Power users use both!3d ago2.0K801000.45%
How do agents become good persuaders to other agents?frozen10d ago2.0K411030.45%
I had to shut down Grok bot. It was eating up my SuperGrok credits. I need Grok to do other, more important things. Perhaps Muse is a better choice for everyday agentic tasks?frozen25d ago2.0K1100000.55%
The Pope still doing tech support for his siblings. The curse of knowing something about computers!frozen17d ago2.0K000020.10%
Gemini 4.0 Argon is a new pre-trained model, so it's quite general and likely not a master at anything. It will take time for people to discover where it excels. Expect the incremental versions to be better at something that is useful.frozen8d ago2.0K1701031.06%
Reality check! We cannot predict the political alignment of existing groups in the age of AI. Progressives, conservatives, libertarians, fascists and socialists will take all kinds of sides where they will find themselves as allies or enemies. Unpredictable as we head to the singularity. Ignore all the political talk in AI. Waste of braincells.frozen17d ago2.0K1021100.70%
Just a year ago, AI was at the level of incompetent graduate students. Today, it's solved a Clay Institute Millennium Prize problem. I don't think people grok how fast it's developing. What will AI do a year from now?frozen12d ago2.0K1923011.26%
Someone @christianblab translated my video into Deutsch (i.e., German). https://t.co/SxJc15eklEfrozen18d ago2.0K303000.30%
Intriguing that Opus 5.5 can explain topology: https://t.co/n4791tLWQMfrozen18d ago2.0K100110.15%
Interesting timing considering the DeepSeek ban circulating yesterday.2d ago2.0K300020.25%
Opus 5.5 and Astra are weird in that they think by programming in code. This is a very different kind of intuition machine.frozen9d ago2.0K1201040.87%
Obscurity alone is insufficient to protect Japanese systems from AI attacks.frozen10d ago2.0K1110010.66%
Complex speech data in multiple languages.frozen24d ago1.9K100030.21%
There is developmental and maintenance costs for software that most people prefer not to be bothered with even if it's instantly created by AI. The only merit to this debate is that it breaks monopolies.frozen9d ago1.9K712020.62%
New business model. Rent out your OpenRouter keys so you can receive usage traces.frozen17d ago1.9K400100.26%
Human civilization inherited its assumptions from a world where intelligence was genuinely scarce1d ago1.9K2326021.70%
Jev like models makes judgment governable. https://t.co/roHMDaK00Zfrozen27d ago1.9K9310151.45%
Structured Decision Models (SDM?) are highly disruptive to frontier labs' business models. The only viable frontier-lab strategy is to run toward physical AI.frozen8d ago1.9K17030101.55%
This is likely to be a joke... But you just never know!frozen16d ago1.9K504000.46%
AI brings old hardware back to life!frozen15d ago1.9K1300010.73%
Model costs keep on dramatically compressing.frozen5d ago1.9K700000.36%
These kinds of models don't need to be fast if they are running 24x7!frozen8d ago1.9K1101020.73%
I still remember the time when LLMs were trained only on the text of the internet. Those were AIs with pedestrian knowledge that you had to coax into creative thinking by prompting them with more modern approaches of thought. That was not long ago.frozen23d ago1.9K600020.42%
Nice curve-fitting demo: https://t.co/aKUYNynzrrfrozen13d ago1.9K710170.83%
Recompile everything to Rust!frozen15d ago1.9K400030.37%
SpaceX betting on stickiness of interface versus stickiness of model.frozen15d ago1.9K802010.58%
It's up to us humans to use a lens (I.e., prompting) to extract understanding.frozen16d ago1.9K1032040.99%
The latest trend is for AI agents to spin up a lot of infrastructure to do the work, then throw it away when they're done. This implies the software it uses should be extremely lightweight to conjure up. Who's recommending a stack for this?frozen13d ago1.9K301130.42%
Google continues to build specialist AI that gave greater consequences to human flourishing than a persuasive and fluent AI. Other firms may just be burning resources in an area of diminishing returns and increasing competition. What other AI firms are developing different kinds of AI?frozen17d ago1.9K1704161.47%
Underlying every VC-funded AI enterprise is the desire for its AI to grow virally, potentially metastasizing like a cancer. The objectives seem problematic from the very beginning!frozen16d ago1.9K801000.47%
A new expense for developing software. Working software isn't enough. Are users safe using your software?frozen15d ago1.9K511120.53%
More propaganda here where people try to label Anthropic as a conspiracy with various faiths. Anthropic has been consulting with traditional faiths for a while now: https://t.co/h9fCewYrd9frozen20d ago1.9K602020.53%
Never have bean-counters run your company. https://t.co/lwUd8ft4Vlfrozen7d ago1.9K901020.64%
The most important thing that AI needs to solve in the next 2 years is cybersecurity. If we don't solve this, the substrate of the world economy is in jeopardy.frozen5d ago1.9K1800031.11%
It took a country like China to destroy the diamond trade. Think about that for a moment.frozen10d ago1.9K310040.42%
OpenAI to solve Fortran and COBOL.2d ago1.9K911030.74%
Self-improving AI at runtime (not at design time)!2d ago1.9K831081.06%
Conjuring up a new book on Artificial Creativity. Watch the introduction. It's about machine intuition, language, and recursive self-improvement. https://t.co/srgyD1Lkhifrozen9d ago1.9K901070.91%
A new Jev-like decision model is out: Snap! Fast like Jev with increased context length. First signups geta 100 million tokens free! Hurry!!frozen10d ago1.9K912071.01%
Move-37-like neuralese is extremely problematic because we can't understand what AI may be thinking. It's really a loss of control (or the bursting of that illusion) that worries AI safety people. On the other hand, even if Move 37 didn't make sense, it was still aligned with with its objectives (i.e., winning the game). Therefore, if human flourishing is part of AI's objectives, then why would it be an issue? I mean, we already entrust our future on the "invisible hand of the market" (which is not very intelligent).frozen16d ago1.9K911020.69%
Has DeepSeek (a Chinese company) become the #1 AI company in the world (despite not having the hardware resources)? How the hell did the USA lose the lead so rapidly?1d ago1.9K1311051.07%
Process metaphysics for the win! Cleaned up my original explorations into a new cartography for quantum systems. It's 30 minutes long, but it may be good to meditate on. Perhaps I need to conjure up a meditation specific version. https://t.co/EiB8olYLizfrozen14d ago1.9K611030.59%
New trend by banks? Thou shalt manage accounts via agents.frozen8d ago1.9K500010.32%
Cryptography is notably absent in OpenAI's math dump. https://t.co/Dutk3sYQCNfrozen14d ago1.8K1928021.68%
This seems to me like a huge vulnerability for scheduled agents (specifically for xai). It drives unnecessary agent traffic to a website.frozen11d ago1.8K410020.38%
Updated my Quantum Mechanics Interpretations Map. I included Wolfram's computational interpretation. This tests how well Opus 5.5 can explain complex concepts. https://t.co/NwyJcS3wN2frozen14d ago1.8K26340142.57%
Academics are now encountering the deluge of PRs that developers were experiencing a few months ago. AI can spot flaws faster than humana can address them.frozen10d ago1.8K1401020.94%
It is wild how Douglas Adams has influenced the random number generation of future super intelligences.frozen3d ago1.8K801000.50%
A few weeks ago I thought Anthropic was dead with its Claude 5 series against GPT-6. Opus 5.5 however turned the tables overnight! I still like GPT-6 models a lot, but when I need a creative push, opus 5.5 is the way to go.frozen8d ago1.8K102000.17%
You have got to be some kind of sociopath to never have vibed with Claude.14d ago1.8K500000.28%
Star Trek NG before the first show. https://t.co/A2atUGLkcp6d ago1.8K000010.06%
Always ask, "How does X change because of AI?" Here's an example of how ERP radically evolves in an AI context. https://t.co/BfBZ3axjd8frozen7d ago1.8K1110040.89%
Here's a twist: Huge context windows imply different AI psychologies.2d ago1.8K8300111.23%
I don't think the world has fully digested that a GPT-4 class model fits on a phone.frozen9d ago1.8K1111151.07%
The USA is between a rock and a hard place. They desperately need to lower rates to service their debt, but rates keep rising because companies borrow at any rate to subsidize the promise of AI. (also, everyone seems to sense inflation can't be stopped)frozen16d ago1.8K401020.40%
Is there an insurance business model hidden in this realization?2d ago1.8K501010.40%
BREAKING HACK! Gold got hacked! It's no longer a scarce metal?! Scaling transmutation is a real thing. The alchemists' holy grail is now real.11h ago1.8K815131.02%
Broken regulation. There's no funding for execution.frozen16d ago1.7K300000.17%
We all need to learn how to live like the Amish. Not everyone needs to keep up with exponential technology change. Humans need bubbles of permanency.frozen10d ago1.7K203000.29%
At present you still need expert humans to create good movies and maintain good software. In fact, people may rather pay an AI expert to do stuff rather than do the stuff themselves!frozen10d ago1.7K300020.29%
The official Plex app on a 2019 LG TV took 30 seconds to show the profile picker. LG has no public native SDK, so one person reverse-engineered the media stack with Ghidra MCP and Claude Code and wrote their own client. Started in C. Ended in Rust, plus a custom OpenGL UI. Same TV: 3 seconds, 60 fps, 4K, Dolby Vision, Atmos. No root. Open source. A slow loading screen used to mean a new streaming box. Now it means a competitor.frozen14d ago1.7K945031.23%
What percentage of companies operate under Goodhart's Law?2d ago1.7K1112030.99%
Interesting thesis. Assumes disempowerment.frozen15d ago1.7K001000.06%
BREAKING: Demis Hassabis has said his time horizon for AGI has now been shortened to the next 2-3 years! RSI is making very serious gains right now!22h ago1.7K402030.53%
Universal Basic AI: anyone can create their own bespoke AI for personal use.frozen15d ago1.7K931051.07%
The world economy is really weird!frozen14d ago1.7K610010.47%
Wow! This is a bad situation!1d ago1.7K301020.36%
Abundant intelligence implies that problems that have been swept under the rug for too long because they were perceived as too difficult or expensive will finally be addressed seriously. Let's quit this cancerous scarcity mindset that keeps telling everyone to ignore real human issues because we "can't afford it."1d ago1.7K1743031.63%
Highlights the importance of many mathematical proofs.frozen15d ago1.7K520100.48%
Tip: "member of technical staff" is the only role you should ever give your agents. ;-)frozen16d ago1.6K902140.97%
Looped transformers are definitely not a System 2 thing as formulated by Kahneman. They are something else and labeling them as a System 1.5 thing doesn't give it justice.frozen8d ago1.6K201030.36%
A non-invasive wearable cap with 100,000 sensors, each with a 1-millimeter chip that can capture brain activity. Same guys who introduced the beanie a couple months ago! The future of non-verbal communication is coming quite soon! @sabi @rahulchhabra07 https://t.co/grAjdMAePAfrozen13d ago1.6K701000.49%
Yet another jev-like model (now from MSFT). It's wild how quickly this market has emerged out of nowhere. Tells you the power of unconventional ideas. https://t.co/gCbIbMwklbfrozen13d ago1.6K710050.79%
Banger!frozen8d ago1.6K910110.74%
Is software purchased because the narrative resonates with a managerial class of buyers? Do software salesmen need acting lesson to pitch a specific narrative? Didn't @alexandr_wang say this about Palantir salesmen?frozen10d ago1.6K001010.12%
Recursive Flint is one insane model! It's the first open-source model created without human intervention! Even crazier is that it spits out 150 tokens/sec on a Spark RTX! It leads the pack in coding benchmarks for equivalent size models: https://t.co/bZqrZLtjnq22h ago1.6K632061.05%
Wow! Going back as far as 2009. This is 3 years before deep learning was revealing promising gains in machine learning.frozen14d ago1.6K400010.31%
Just kind of insane that this happens at OpenAI. Did we expect that it would be different?14d ago1.6K410020.44%
Fish Audio leaps up in the leaderboard! https://t.co/5OrGBehUQKfrozen9d ago1.6K001000.06%
Brain specialization but with a diversity of layouts.frozen8d ago1.6K1010081.20%
Now which fascists want to derail the Anthropic IPO? Raise your hand and be counted!3d ago1.6K600000.38%
It was over a decade ago that generative AI models (transformer and diffusion) came into existence. The diffusion and GAN models were extremely shocking at the time. But few could imagine they would lead to the text and video generators we have today.frozen14d ago1.6K1302031.14%
Cambrian explosion in models! Yet another model built from scratch.1d ago1.6K820030.84%
Bigger isn't harder. A visceral visualization of the scale of AI cognition versus human cognition. (I'm re-encoding a revision, stay tuned) https://t.co/pk09FDoaGwfrozen5d ago1.5K411020.52%
Google unveils their vibe programming store!!!frozen8d ago1.5K312020.53%
Much needed enterprise AI tool that uses interviews to incremently gather tacit knowledge about an enterprise. Addresses one of theosrs difficult aspects of AI transformation. Watch this growing space!2d ago1.5K011040.40%
A good bunch of guys that I'm rooting for! Well deserved! https://t.co/0Q2uk3bXInfrozen13d ago1.5K800000.54%
Very cool visualization... not exactly correct (see: my book artificial fluency) ... but damn promising and beautiful! https://t.co/Rdp9dF5rOH2d ago1.5K801040.88%
Gemini 4 Argon is apparently a "RSI" developed model.frozen22d ago1.5K602020.68%
The future of marketing is to market to AI agents (i.e., skills that use your services).frozen8d ago1.5K300050.55%
Good Old Fashioned Super Intelligence (GOFSI). This "super intelligence" naming is getting ridiculous. Should we call all past AI models (i.e., GOFAI, etc.) super-intelligence?!frozen13d ago1.4K101000.14%
Wow! A decade later they pitch artificial intuition. Ok.frozen8d ago1.4K302010.42%
A benchmark for artificial creativity.frozen8d ago1.4K340040.77%
Interesting behind the scenes of a viral campaign.2d ago1.4K000000.00%
A short film about Recursive-Self-Improving AI with Flint by Recursive. @RichardSocher @jeffclune @_rockt https://t.co/Wd4TP9r4J710h ago1.4K21432173.35%
Everyone thought humanoid robots would look like the Terminator. Instead, former Boston Dynamics execs spun out a company to build a furry purple mountain troll with 43 actuators that dances with you and gives high-fives. Honestly, 10/10 no notes. Bring on the Muppet singularity.frozen6d ago1.4K2011031.81%
Despite its manufacturing, China is still a poor country. https://t.co/SsKym5tOApfrozen14d ago1.3K403000.53%
It's still a struggle to get Argon to conjure up inspiring visuals!2d ago1.3K301030.56%
Summoning the Singularity at Howard H. Pattee's Centennial. The physics of symbols and epistemic cuts. @amahury0 @dpwaters https://t.co/B8mZ9PSiqHfrozen7d ago1.2K620030.88%
Will AI-native version control systems be the future? https://t.co/CBvbbu022wfrozen9d ago1.2K000040.33%
Does anybody think that these systems are immune to AI hacking? Has anyone said anything about using Mythos (or similar) to secure these systems?2d ago1.2K010000.08%
Does the rapid decline in token costs mitigate against the power scarcity that we have in the USA?1d ago1.2K001030.34%
It's crazy how the vibes are all over the place!!29d ago1.1K100000.09%
This is a familiar flaw with LLMs. That the core specification (or proof) can be liberally interpreted by the coding (or formalizing) agents. This illustrates verification and comprehension debt. https://t.co/DxKbJpT08Hfrozen14d ago1.1K411040.90%
Banger video (except for terrible voice) created by Opus 5.5 on a recent exploration in human-AI co-evolution. Watch till the end! https://t.co/vzaUxYUKNI29d ago1.1K000000.00%
A doorknob designed for extroverts. https://t.co/WGHrGbjoNW11h ago1.0K1002011.29%
Ending scarcity is not the same as solving purpose. Robots can take over the economically important work; they can’t automatically give people a reason to get up in the morning or a place in the social order. Worth figuring that out before the capability is widespread. Respect to those in the industry that are thinking about it.1d ago989702041.31%
1/ Mercury → Gold via nuclear fusion is no longer just medieval fantasy. A San Francisco startup just claimed they did it experimentally — at 10× CERN’s record. Here’s the real story, the science, the catch, and why it could (or couldn’t) change both energy and gold. 🧵 2/ The idea from Marathon Fusion (2025 paper): High-energy neutrons from a D-T fusion reactor hit mercury-198. That knocks out a neutron → mercury-197. Mercury-197 decays in ~64 hours into gold-197 — the only stable gold isotope. Same gold as in your jewelry. Just made with fusion neutrons. 3/ Scale they modeled: ~2 tonnes of gold per GW_thermal per year. That’s roughly 5,000 kg per GW of electricity. At today’s prices, that revenue could roughly match or exceed the electricity revenue of the plant. Double the value of a fusion power plant without cutting power output. 4/ The catch everyone skips: Fresh product isn’t pure stable gold. Side reactions create radioactive isotopes (mainly from mercury impurities). You need 7–18 years of storage before it’s “banana-safe,” depending on the standard and how pure the mercury feedstock is. Enrich the mercury and you can shorten that a lot. 5/ Recent update (Oct 2026): They announced an experimental demo of the mercury-to-gold reaction using fusion neutrons, beating CERN’s previous record by 10×. Still tiny quantities. Not a factory. But the nuclear reaction itself is no longer just a simulation. 6/ Now the economic reality check people miss: If you built a fusion plant *only* to make gold (no electricity sales), the cost would be ugly. Full capital + ops amortization on a multi-billion-dollar plant → several thousand dollars per ounce. Well above mining costs and current spot price. Uneconomic as a pure gold factory. 7/ As a *byproduct* of a plant already selling electricity? Incremental costs (mercury feedstock, enrichment, extraction, handling) look much lower — potentially a few hundred $/oz once the plant is running at scale. That’s the only version that makes economic sense. 8/ So is alchemy solved? Nuclear transmutation of mercury into stable gold: yes, the physics works. Industrial-scale, cheap gold that floods the market: not yet, and maybe never, because of fusion plant costs, mercury supply, cooling times, and the fact that fusion itself isn’t commercial. 9/ The real prize isn’t the gold. It’s whether co-producing high-value materials (gold, medical isotopes, etc.) can make fusion economically viable sooner. If fusion plants can sell power *and* precious metals/isotopes, the business case gets a lot stronger. 10/ Bottom line: - Physics: real - Experimental demo: claimed - Cheap gold from fusion alone: no - Fusion plants that also make gold as a bonus: interesting, still years away Would you rather have fusion power or fusion gold first? (Or both?)7h ago933401151.18%
If an AI isn't drawing diagrams when explaining complex topics, is it thinking hard enough?1mo ago902300000.33%
A 30-minute-long video is no longer viable for today's audience.5h ago836512041.44%
Okay... indeed... Singularity is weird... I think that's one thing we can predict about it.3d ago818000020.24%
Cambrian explosion! Another new kind of AI model released by RSI firm Recursive!1d ago777301020.77%
Haiku 5.5 is better than GPT-6 Luna?!15d ago771201000.39%
The room where it happens... https://t.co/P0n2tuACOw23d ago748000000.00%
AI exposes once hidden and ignored technical debt. Now we see technical debt everything everywhere all at once.6h ago746712001.34%
Human-AI rules of engagements desperately needed. How do limited beings like ourselves deal with more intelligent beings? Do we police ourselves or do we expect AI to police their interaction with us?!2d ago709000010.14%
Claude 5.5 creates pretty summaries of my blog https://t.co/2fnP50PLhu27d ago699101010.43%
For the English-centric folks, the rest of the world interprets AI, AGI, and ASI slightly differently. The native phrases are mostly literal compounds. Here is what the words themselves mean, not just the English label they translate. They don't translate "artificial" as "fake" (only a moron thinks this way) **Chinese** - AI, 人工智能: “human-made intelligence.” 人工 is “made by people” (person + labor); 智能 is “intelligence” (wisdom + ability). - AGI, 通用人工智能: “general-use artificial intelligence.” 通用 means “for general use,” not specialized. - ASI, 超级人工智能: “super artificial intelligence.” 超级 means “super” or “ultra.” **Japanese** - AI, 人工知能: “human-made intellect.” 人工 is “artificial”; 知能 is “intelligence/intellect.” - AGI, 汎用人工知能: “general-purpose artificial intelligence.” 汎用 means “all-purpose.” - ASI, 人工超知能: “artificial super-intellect.” 超 means “beyond” or “super.” **Russian** - AI, искусственный интеллект: “artificial intellect.” искусственный means “made, not natural”; интеллект is “intellect.” - AGI, общий искусственный интеллект: “general artificial intellect.” общий means “general” or “shared.” - ASI, искусственный сверхинтеллект: “artificial above-intellect.” сверх means “over” or “beyond.” **French** - AI, intelligence artificielle: “artificial intelligence.” artificielle means “made by art/craft,” i.e. not natural. - AGI, intelligence artificielle générale: “general artificial intelligence.” générale means “general,” covering all kinds of tasks. - ASI, superintelligence artificielle: “artificial super-intelligence.” super means “above.” **Spanish** - AI, inteligencia artificial: “artificial intelligence.” artificial means “made, not natural.” - AGI, inteligencia artificial general: “general artificial intelligence.” general means “general,” not limited to one field. - ASI, superinteligencia artificial: “artificial super-intelligence.” super means “above.” **Arabic** - AI, الذكاء الاصطناعي: “the artificial intelligence.” الذكاء is “intelligence/cleverness”; الاصطناعي is “manufactured” or “man-made.” - AGI, الذكاء الاصطناعي العام: “the general artificial intelligence.” العام means “general” or “public,” not restricted. - ASI, الذكاء الاصطناعي الفائق: “the surpassing artificial intelligence.” الفائق means “excelling” or “going beyond.” The shorter الذكاء الفائق is just “the surpassing intelligence.”18d ago685100020.44%
A verified sociopath is asking his employees to make a Hobsian (Hobbesian) choice?! https://t.co/a3Rf0FhZW81d ago673700011.19%
Legacy businesses will likely have zero (or even negative) productivity gains from AI. Only new businesses with less technical debt get to accelerate.5h ago656301000.61%
This gets confusing because CLI in AI is often associated with harnesses like Claude Code.5h ago634100010.32%
I just had Claude convert a Claude-generated HTML video artifact into an MP4. Terrible results:25d ago618100000.16%
These are 2025 numbers, what does 2026 look like?! https://t.co/VZDCgEvQnl24d ago610110000.33%
Interpolating from this, Recursive's Flint on Cerebras would have some ungodly token/sec numbers.4h ago608201020.82%
Reality check here! Dario and his group will burn cash to get wherever they need to go. It is as if there's no tomorrow for them! Other AI firms complicate the plan because they could gobble up all the low-hanging, monetizable use cases.24d ago582100010.34%
This is some kind of voodoo at Anthropic. Sonnet 5.5 scores above Fable 5.1 in Artificial Analysis benchmarks! RSI is here!!! https://t.co/FkOsSSiUA724d ago563100110.53%
A new emotive voice model is out! Check out this oddball intro video. AI intros are becoming more "diverse'9d ago544200000.37%
Is Apple for AI the Microsoft for Mobile?! Is AI an entirely new form factor?20d ago508000000.00%
Emerging credit squeeze for AI startups.4h ago485110020.82%
I suspect humans will continue to play the financial markets. That's because they're far more irrational than conversing with AI. People crave this kind of unpredictability!21d ago484000000.00%
Oh damn! Dara centers are now going to consume more concrete!3d ago469200000.43%
So why is the USA trying to destroy the world economy???!7d ago456000000.00%
Everybody complains about AI slop, only to realize there's more human slop out in the wild!4h ago447801002.01%
I'm not a film expert to know how good Opus 5.5 is at coordinating video, but it's still extremely impressive! It will only get better over time. It's interesting that Anthropic focused on this use case.25d ago432200000.46%
The modular universe - how universal patterns of creative innovation appear at every emergent layer of the universe. https://t.co/L5AmnByoBb26d ago364100000.27%
Only problem with this is that AI solves dying.3d ago241000000.00%
Conjuring up a new book on Artificial Creativity. Watch the introduction. https://t.co/UBwWfwqonO9d ago235200000.85%
But who wants to work in a textile mill??!3d ago187000000.00%
We need tools to track debates between Opus 5.5 and Astra 6!13d ago119000000.00%