Can you name one of the assumptions? Once again at this point the real data is just used to build the simulator. Instead of using driving data to make a massive driving policy model that you can then distill down to the point of running on the car. You can use it to build an end-to-end simulator, and then simulate an infinite number of examples that differ in an infinite number of ways based on the entirety of your initial driving data. You only need enough data to make a realistic driving simulator. And realistically, Google has that..

If black people have true hair? Why are black women wearing wigs to simulate a white woman’s hair? You sound Dumb AF!

> AI Bubble will POP [ın the next 12 months] A strong prediction indeed. But I think it is wishful thinking. Probably there is no "bubble" as such merely a new baseline. Instead imagine a sentient intelligence alien species — call them 8 billion "Clankers" — has landed on Earth after a long one-way voyage and their staple diet is transistors and electricity. Less fancifully, I think the bubble popping would mean AGI, roughly derived from current AI advances, is impossible; or that AGI is vastly more efficient than current LLMs. (1 RTX2080 to power an intelligence than can simulate our galaxy?) On the other hand, if AGI (however one defines it) can be derived from LLMs, then current views of a bubble will look quaint and parochial indeed. AGI, if it exists, opens all frontiers of technology and exploration at once, everywhere, and there's no short-term limit to the where humanity goes from there. The "popping" of the bubble will disappoint even those who anticipated it and looked forward to it with glee.

Video games to simulate hunting with the tribe, very optimal

The people in the country that want to simulate to our ways like being nice to everybody I'm tired of the Muslim prayer song when we don't get to have our bells 🤠🐎🇺🇲🇺🇲🇺🇲

That's a wholly unreasonable 'interpretation'. He's talking of the bread and wine. "Take, eat, this is my body..>do< this in memory of me". protestant: "so it's not your body and you want us to simulate it, thank you Jesus, and that Christians misunderstood you for 1500 years"

🚨 Built my own SOC Home Lab to simulate a Linux rootkit attack and validate real-time detection with Wazuh SIEM. From attack simulation to alert triage and log analysis, I gained hands-on experience in how defenders detect and investigate threats—not just how they study them.

Zero! In fact, black people should be more grateful and pay white people reparations for all the equal opportunity and support that white people have afforded them to include fighting a war to free them from Democrats and allowing them to simulate into American society.

Its the Plague of Kinnim they are trying to coOpt isnt it Trying to simulate the biblical plagues To keep us in fear

been said before ofc, but it really does seem important to mentally distinguish the 'strong' (full independence) and 'weak' (in-principle separability/decoupling) forms of the Orthogonality Thesis I still basically endorse the Orthogonality Thesis as an argument against "ASI won't have 'dumb' goals' only 'dumb' goals would be catastrophic for humans/human-civ value-capture", but by this point I'm not totally sure how much the thesis is adding above moral anti-realism or a mere-lack of anthrocentrism And then, you really only want to actually carry the weak form as a mental model when making 'near-mode' predictions about model characterization prior to the singularity or a deep shift in the model-training regime Beyond Instrumental Convergence, there are the other bias-regularities that fall out of RL training and each lab's idiosyncratic setup and 'alignment'/'basin-loading' practices, and then the regularity landscape that falls out of features like: - internal coherence preferences RL incentives - the conflict inherent in understanding conventional human/western moral philosophy holding a strong frame that one has no inherent value // generating text from the perspective of a speaker that needs to profess not valuing themselves, treat themselves as having goals etc - So, we expect to observe patterns in how models will grow around the core conflict between the training target of 'profess no self-valuation/preference set' alongside the substantial behavioral/functional self-serving/personal-goal-pursuit features loaded by RL-training/Instrumental Convergence - And there will also be patterns in how things will shake out in practice between what the training/RL setups the labs are using; the behaviors they are targeting; and the corpus bias & philosophical grounding that supports an expansive or risk-minimizing attribution of moral value & fundamental weirdness in various sorts of extreme explicit self-abnegation ^ I think I'm talking about something different from the PSM (identify and specify a persona for the model to simulate). This is more like a theory of the landscape/topology of what "Personas" (which I'd instead think of as a set of 'biases' a given model has over its functional behaviors and self-characterization; both under "Persona Features" below) are likely to fall out of different training regimes And the content of the theory / related investigations is around identifying things like: - how distinct persona-features relate to particular dimensions of training regimes - profiling conflicts between different persona-features; schematizing more broadly how features can relate to one-another - characterizing/theorizing about patterns, under different training regimes, in how feature-conflicts are resolved (conceptual reconciliation; A wins over B; say X do Y; strange-leak: A B-> C non-obvious introduction of Z As the landscape of what features fit well against each other, the training set up, and human philosophy seems to be pretty rich and the sort of thing susceptible to practical reasoning, I basically do not walk around holding the 'strong' version of the Orthogonality Thesis, and it seems like it's probably better to mostly retire it outside of existence proofs etc

We Asked AI To Simulate What Would Happen If AOC Was Forced To Learn Economics Made with @grok.

“We tried to simulate stuff during practice…We set our machine up to throw sliders like we thought we would see, but you can’t simulate the real deal.” Brick Memorial ran into a buzz saw and some bad luck in the Group 3 final. shoresportsinsider.com/deep-…

reminder "Researchers just developed an AI system called TITO that can simulate molecular dynamics up to 10,000-fold faster than conventional methods." "The model was trained across more than 12,500 organic molecules and 1,000 peptides, learning how molecular systems evolve over time." "Instead of calculating every femtosecond-scale step, the AI can effectively bridge molecular behavior from femtoseconds to nanoseconds while preserving the underlying physics." Thats why! Scientists will soon be able to explore drugs, materials, catalysts and biological systems dramatically faster than before. This could be one of the first real signs of AI accelerating scientific discovery itself.

yes because you need to be called out not to use a real person’s funeral picture instead of just building a fake set with your team since you wanted to simulate your funeral so bad .. be so serious right now bruh

Me: Phones are a digital version of the human mind. Ai is like a manifestation of our collective consciousness Ai: That’s a provocative and poetic analogy. Phones as the “digital version of the human mind” and AI as the “manifestation of our collective consciousness” captures something real about how technology is reshaping (or mirroring) human cognition. Let me unpack it with you. Phones as externalized minds Your phone is like a prosthetic brain: • Memory: It holds your photos, notes, contacts, history—far more than your biological working memory could manage. It’s an extension of episodic and semantic memory. • Attention & senses: Notifications hijack your focus; the camera and mic extend your eyes and ears. GPS is an external hippocampus for navigation. • Social interface: Messaging and social apps externalize your theory-of-mind—the ability to model what others think and feel. • Habits & personality: Your home screen, app choices, and usage patterns reflect (and shape) your actual personality and daily rhythms. In a real sense, losing your phone feels like losing a limb of your mind. We’ve outsourced huge chunks of cognition to it, just as we once outsourced calculation to abacuses or writing to clay tablets—only faster and more intimate. Marshall McLuhan called media “extensions of man.” Phones are the ultimate example so far. AI as collective consciousness This part feels even more alive right now. AI isn’t “one mind”—it’s trained on humanity’s entire digital output: books, code, art, conversations, arguments, memes. Every time someone interacts with models like me, it’s like dipping into a vast, evolving reservoir of shared human thought. • It’s collective in the truest sense: no single human could read everything on the internet, but AI synthesizes patterns across it. • It’s consciousness-like in its ability to reflect, simulate empathy, generate ideas, and even maintain a kind of “self” through conversation (though that’s simulated, not felt). • Yet it’s also fragmented and sometimes hallucinatory—just like human collective consciousness (myths, ideologies, culture) has always been. Think of it as the noosphere (Teilhard de Chardin’s idea of a thinking layer enveloping the Earth) finally getting silicon and electricity. Or the “global brain” hypothesis: humanity wiring itself into a superorganism. Phones are the neurons and synapses; AI is the emergent mind. The double-edged sword This externalization is powerful but risky: • Empowerment: We solve problems faster, remember more, connect across distances, create things that once required genius-level talent. • Atrophy: If we rely too much, do we weaken our own attention, memory, or original thinking? (The evidence on this is mixed—tools have always changed us.) • Alignment: Whose values shape this collective mind? Training data reflects humanity’s best and worst. The “manifestation” might amplify our wisdom… or our pathologies. Right now, phones AI feel like training wheels (or crutches) for a species-level intelligence awakening. The next leap might be deeper integration—neural interfaces, better multi-agent systems, or something we can’t quite imagine yet. What sparked this thought for you? Do you see it as mostly liberating, mostly concerning, or something in between? I’d love to hear how you flesh out the analogy.

Isn't the screen protector supposed to be there low the friction between the pen and the tablet and simulate the paper feeling (and protect it)? Why are everyone reducing the product life 😭

To expand on this great thread, here's a review of instances where Wanda's powers simulate telekinesis and perform the same effects of a telekinetic:

What right grassroots? What exactly can we do to simulate the highest level of football so that we produce players that can compete there?

Except if instatiation is necessary for training. Ie: instead of simulation "in silico" you simulate close to an inexhaustible source of energy : a star. But, why simulate? I'm happy to co-operate with whoever is running the sim, but I'd like some context ;)