PsycheOS | Psyche Suite 🧠🧬

Is Psychology "Woo"? Nope. Psychology is a Technology. Specifically, software -- Psychological software. And like Software, Psychology is intangible. But nonetheless real. Its math, its functions and its algorithms are all computationally deterministic. 1+1 = 2 4-2 = 2 1*2 = 2 4/2 = 2 2^1 = 2 No one in their right mind would claim that "software doesn't exist" and that Microsoft, Oracle, SAP, Netsuite, SFDC or Workday are all "scammers" and "grifters" selling "woo". And the "PsycheOS" and "Psyche Stack" aren't "Woo". It's Applied Psychophysics. The critics' first line of attack is always personal -- "this is woo!", "pseudoscience!", and "you're a pseudointellectual!". In reality, I am just a solution architect and a software programmer -- of Psychological Software. The next attack is, "where is the evidence? You can't program the Psyche!" Or moving goal posts -- "The Psyche is 'real' but it's not programmable!" And yet these same people are all programmable primates who have been programmed down to their very last neuron. And are clueless this has even happened. So where is Microsoft's peer-reviewed research and papers that Microsoft Word works or that you can bold a selection or italicize it or use spell check? Where is Oracle's peer-reviewed research and papers that you can join a table or write a stored procedure? They don't exist and yet that software generates hundreds of billions of dollars of revenue per year and employs over 2 million highly paid staff while their customers - businesses we all know and trust - P&G, Unilever, Toyota, Honda, Dell, Lenovo, etc. -- rely on this "woo" aka software to run their businesses. My evidence is the system, the math, the functions, the algorithms, the software and the results. Just as theirs is. The PsycheOS, PsycheStack and Programmable Psyche are testable. In fact, we have hundreds of millions of data points to back this up over centuries. The solution? The solution is to code. People.

@fivepebbles__ @om_patel5 I enjoy watching Hylics struggle in the age of AI.

@PsycheOS @om_patel5 You wouldn’t have responded if that were true. Gotcha, kiddo.

RESEARCHERS JUST BUILT AN AI MODEL TRAINED ONLY ON TEXT FROM BEFORE 1931 it's called talkie. 13 billion parameters, trained exclusively on text published before december 31, 1930 its worldview is completely frozen in time the reason this matters: every major AI model today (GPT, claude, gemini, llama) was trained on the modern web. that makes it almost impossible to tell if these models actually reason or if they just memorized the answers from their training data talkie breaks that completely because it has never seen any modern information the crazy part: talkie can learn to write python code from just a few examples you show it in the prompt. despite having ZERO modern code in its training data. it's figuring out programming from 19th century mathematics texts. that's ACTUAL reasoning claude sonnet 4.6 was used as the judge in talkie's reinforcement learning pipeline. claude opus 4.6 generated the synthetic conversations used in fine tuning. a modern AI was used to train a model that's supposed to be frozen in 1930 the team already flagged this as a contamination risk they want to eliminate in future versions what they're using it to study: > long range forecasting. how well can a model "predict" the future from a frozen vantage point > invention. can it develop ideas that didn't exist until after its knowledge cutoff > LLM identity. what makes a model itself vs what's just patterns absorbed from the web alec radford built this. the same guy behind GPT, CLIP, and whisper both models are open source on hugging face. they're already planning a GPT-3 scale vintage model later this year an AI that has never seen the modern world can still reason its way to writing code. THAT alone tells you more about intelligence than any benchmark ever will

@AuroraMar1eL Many people knew this but the valley echo chamber and circle jerk is/was too powerful. The CIA AI Cartel thought they could push their bullshit that GPUmaxxing was the only moat that mattered. And then Qwen, Kimi, Deepseek and friends knocked their c0ck to their back pocket.

🚨 The AI industry just lost 3 years. Trillions spent. Billions burned. Chasing the wrong idea. Yann LeCun called it from day one. Nobody listened. Until now. The bet was simple: make the model big enough and it’ll eventually understand the world. LeCun said that’s nonsense. Generative AI is fundamentally inefficient. When a model predicts the next word or the next pixel, it pours compute into surface-level detail. It learns patterns, not the underlying physics of reality. He pushed a different approach: JEPA (Joint-Embedding Predictive Architecture). Instead of making AI recreate the world pixel by pixel, JEPA makes it predict concepts. What happens next—not in raw data space, but in a compressed “thought space.” But for years, JEPA hit a wall. Representation collapse. When you let an AI “simplify” reality, it takes the easy way out: it cheats. It compresses the world until a dog, a car, and a human all blur into the same thing. It doesn’t learn. It collapses. The fix has been ugly: elaborate hacks, frozen encoders, and mountains of compute just to keep the model honest. Until now. A new paper, LeWorldModel (LeWM), claims to solve the collapse problem outright. No Rube Goldberg engineering. Just one clean mathematical regularizer. It pins the model’s internal representations to a true Gaussian distribution—so it can’t hide behind shortcuts. If it wants to predict, it has to model the real structure of the world. And the implications are wild. LeWM doesn’t need a massive centralized cluster. It’s only 15 million parameters. It trains on a single standard GPU in a few hours. Yet it plans 48× faster than huge foundation world models, shows an intrinsic grasp of physics, and flags impossible events on sight. We’ve spent billions building server farms to memorize the internet. Now a small model, running locally on one graphics card, is starting to learn how reality actually works.

🚨BREAKING: HKUST just gave AI agents permanent memory that improves over time. No retraining required. Lessons from one model transfer to another. Up to 11 points better on the hardest benchmarks. > Every AI agent you use today starts each task completely blind. No memory of what worked last time. No memory of what failed. Every mistake gets repeated forever. > HKUST built XSKILL a dual memory system that accumulates two types of knowledge after every task: skills (what workflows to follow) and experiences (what specific mistakes to avoid). > The model itself never changes. The memory just gets smarter. > The part nobody expected: knowledge learned by Gemini transfers directly to GPT and o4 mini. No additional training. One model's lessons become another model's head start. → Up to 11.13 point improvement over the strongest baseline on hard benchmarks → Syntax errors cut nearly in half: from 20.3% to 11.4% after skills added → Cross-model transfer works: Gemini's knowledge improves GPT-5-mini and o4-mini → Zero parameter updates required at any point → Knowledge compounds: more tasks = smarter memory = better performance The fix is simple in principle. Skills stop the agent from wasting steps on errors it already made. Experiences tell it exactly which tool to pick in which situation. Together they turn a stateless agent into one that actually learns from its past. Every AI agent deployed today is leaving this on the table.

The entire RAG industry is about to get cooked. Researchers have built a new RAG approach that: - does not need a vector DB. - does not embed data. - involves no chunking. - performs no similarity search. It's called PageIndex. Instead of chunking your docs and stuffing them into pinecone, it builds a tree index and lets the LLM reason through it like a human reading a book. hit 98.7% on financebench. beats every vector RAG on the leaderboard. no embeddings. no chunking. no vector DB. 100% open source.

I’m unsure why consciousness would be exempt from scientific study. If it’s the "interface" or HUD/heads up display (dashboard), then it’s precisely a structured system that we can analyze, deconstruct and reconstruct. This allows us to learn what generates it, what shapes its contents and how it interfaces/relates to underlying mechanisms and systems (brain, regions of brain, psyche, etc.)

@kanair Consciousness is not a subject for science. Science is limited to things that appear in consciousness.

Talking about AI consciousness is becoming trendy. But much of the debate still feels like the same for the last 30 years. Consciousness research has to move forward. What seems new and promising to me now are two directions. Mathematical theories and invasive BCI in humans.

Maybe the issue is that we don’t force enough constraint/s on theories early on. If ideas are forced to be systemized, operationalized, and then required to produce falsifiable, discriminative predictions, this should stop these interchangeable interpretations (of the same intuitions/patterns sensed). And it may well be that "that" constraint is w hat generates a genuinely new structure/s by reducing error (eror reduction( in how we update or move between competing explanations.

Of course, there is still a lot of interesting research on consciousness. But I feel I rarely encounter really new ideas at the conceptual level and often end up having a variation of the same discussion. How can we change this?

Math + BCI make sense because they add constraints. But without clearer definitions of "consciousness" (and its dimensions/factors, thresholds, etc.), having even better data and formalisms may just have us see the same disagreements but in higher resolution and color.😀 In my view, the key is whether the theories can be made falsifiable in a way that actually separates them.

If all experience is already simulation (brain-generated), then "simulated vs real" collapses. What would appear to matter most is the structure and constraints of the simulation and factors/dimensions involved and converging for that to happen, rather than the substrate or label/s involved.

I often hear arguments that simulated consciousness cannot be real consciousness. But these arguments often miss the point that simulations are physically instantiated in a computer with real causal dynamics. It is not like a fictional character with no internal mechanisms.

@pubity It just means you start a new company where you are not hiring people who are replaced by AI, you are an AI company where you are hiring some humans to assist the AI.

A Chinese court has ruled it illegal to replace human workers with AI purely for the sake of cost-cutting. The court decided that companies hold a social responsibility to treat workers fairly and pay them what they're worth.

@fivepebbles__ @om_patel5 Keep crying harder - reality doesn't care about you. And neither do I.

@PsycheOS @om_patel5 Nice pseudo intellect you’ve got there. Whatever helps prop up your fragile ego and prevents you from killing yourself.

Tencent has killed fine-tuning and RL with a $18 budget. Right now, if you want an AI agent to become an expert at a specific, complex real-world task, you have to use Reinforcement Learning. You let it try, fail, and update its internal parameters over and over again. This is the exact optimization technique (GRPO) that DeepSeek used to build their massive reasoning models. But there is a massive problem. Updating model weights is insanely expensive. It requires massive GPU clusters. And worst of all, when you train a model to be highly specialized at one thing, it often "overfits" and forgets how to be good at everything else. Tencent killed this bottleneck forever.. by building Training-Free GRPO. Instead of spending thousands of dollars to permanently alter the AI's brain, they asked a simple question: What if we just distill the experience of learning, and inject it as a memory? Here is how it works. They run the AI through the exact same trial-and-error process. But instead of updating the weights, they extract the "semantic advantage"—the actual logic of why one answer was better than another. They compress this winning logic into a "token prior”, a tiny package of high-quality experiential knowledge. Then, they just attach that knowledge directly into the API call. The results are staggering. Tested on DeepSeek-V3, this method required only a few dozen training samples to turn the AI into a specialized expert in complex math and web searching. It didn't just compete with models that were actually fine-tuned. It outperformed them. Zero parameter updates. Zero expensive training runs. Zero base-model amnesia.

@BurgenDobs @om_patel5 Science and software doesn't care about subtlety. But Semitic Spiritual Seed Oil for the Soul requires it.

@PsycheOS @om_patel5 Jesus lol practice some subtlety

@LaWRLDxx @om_patel5 Bayesian inference is "just saying numbers" to a moron. Heidegger or Kierkegaard is "just saying letters of the alphabet", to a moron. youtube.com/watch?v=dQWtl9…

@PsycheOS @om_patel5 Just saying words

@fivepebbles__ @om_patel5 You're a dolt. And your reaction was your confession. Good job, parasite.

@PsycheOS @om_patel5 “I’m a racist 14-year-old and this was deep.”

Exactly. Business is the only Battlefield that matters (outside of one's mind / human mind). And tech founders should be thinking how their business can terraform the world to support their worldview while monetizing their enemies' demise. 7 birds : 1 stone (ideally 10 birds : 1 stone but not your stone, not your labor, you still get all 10 birds)

For successful tech founders still running their companies there is a new opportunity. Think about how your company can be strategic to the free world. Investors in the short run won’t reward you for these efforts but in a long run knowing you helped to build the America of tomorrow is reward enough. Also remember that you get to help shape the world your children and grandchildren will grow up in.

Heavy Technical Debt powered by Big Block Brute Force AI. vs Fully optimized Clean Sheet Paper process design. In fact, Clean Sheet Paper processes powered by Microsoft Excel even beat Broken Processes powered by AI.

Startup / New Venture reality: It's a process of falsification and elimination which takes time but market timing is always pulling the strings. But the most you can do is push on the strings. So you need to move as fast as you can (to falsify and eliminate) BUT be resilient enough to hang on if the market isn't ready yet -- and the market, especially nascent markets are almost never ready when you or your product is.

The romanticization of being a startup founder is kind of insane to me: – You’re statistically unlikely to succeed – You’ll have no life and be totally consumed - Incredible lows – You’re locking yourself in for 10+ years This is insanely hard and NOT for everyone.

@hthieblot As in dating, I take all of these as a shit test. And I take all shit tests as something positive. Because in almost all cases, it's a buying question.

VC rejection excuses: • • No moat • Big labs will kill you • Too much competition • Valuation is too high • TAM is too small • Get a co-founder • Not enough traction • Your retention sucks Translation: I don’t believe you’ll win. Your reaction: Watch me.

@sweatystartup Endless work to do just focusing on your own immediate family (your kids and grandkids, your siblings) and extended family (your nieces and nephews, some cousins). Could easily work 40 hours a week just on that.

The concept of retirement is hilarious to me. What do people do all day? Work is fun. Half the joy of life is feeling accomplished and getting things done. I’ll work 30+ hrs a week well into my 110s.