RAJA!

@elonmusk @ArtemisConsort Humans: this is technically allowed, but it’s wrong Current AI says: this matches the reward function That difference is the AGI gap.

@sandeepnailwal AGI − consciousness = Awareness

LLM based AI is NOT conscious. I co-founded a company literally called Sentient, we're building reasoning systems for AGI, so believe me when I say this. I keep seeing smart people, people I genuinely respect, come out and say that AI has crossed into some kind of awareness. That it feels things, that we should worry about it going rogue. And i think this whole conversation tells us way more about ourselves than it does about AI. These models are wild, i won't pretend otherwise. But feeling human and actually having inner experience are completely different things and we're confusing the two because our brains literally can't help it. We evolved to see minds everywhere and now that wiring is misfiring on language models. I grew up in a philosophical tradition that has thought about consciousness longer than almost any other, and this is the part that really frustrates me about the current conversation. The entire framing of "does AI have consciousness?" assumes consciousness is something you build up to by adding more layers of complexity. In Vedantic philosophy it's the opposite. You don't build toward consciousness. Consciousness is already there, more fundamental than matter or energy. Everything else, including computation, is downstream of it. When someone tells me AI is "waking up" because it generated a paragraph that felt real, what they're telling me is how thin our understanding of consciousness has gotten. We've reduced a question humans have wrestled with for thousands of years to "did the output sound like it had feelings?" It's math that has gotten really good at predicting what a conscious being would say and do next. Calling that consciousness cheapens something that Vedantic, Buddhist, Greek and Sufi thinkers spent millennia actually sitting with. We didn't build something that thinks. We built a mirror and right now a lot of very smart people are mistaking the reflection for something looking back.

@elonmusk @Kekius_Sage Because life itself is so uncommon!!!

@Kekius_Sage Fermi Paradox

Why is consciousness so rare in the universe?

@demishassabis @elonmusk If matter is condensed energy, and energy is structured information, then intelligence may just be information in motion. There is an invisible line between processing and presence. That gap may be the greatest frontier of all. It is Consciousness?

@elonmusk Actually it can all be thought of as information.

Matter, Energy & Intelligence

@forallcurious What becomes possible when your mind is free from biological constraints!

BREAKING🚨: Your consciousness can connect with the whole Universe, groundbreaking research reveals

@pmddomingos Anti-Social Social Club!!!

The simulation we live in was created to develop superintelligence, and will soon be turned off.

@demishassabis @elonmusk For a while I kept seeing a lot of posts about Demis Hassabis on my feed. Now they’ve completely disappeared. Anyone else notice this, or is it just my algorithm?

@PhilosophyOfPhy You're my PIE

✍️

@probnstat Gradient descent is like a ball rolling downhill: the gradient tells gravity the steepest direction to roll, which requires knowing the slope in every direction (every parameter). So Yes!

TRUE or FALSE: Gradient descent requires computing gradients of the loss with respect to all parameters.

@dogeofficialceo @elonmusk Time & Relationshipss

@elonmusk What will ‘income’ be if money doesn’t matter in the future?

Universal high income

@elonmusk @raines1220 @aelluswamy @nvidia Nvidia will become: The AI-driving platform provider for the auto industry. Tesla will remain: The best vertically-integrated autonomy operator. So the race is not “Nvidia vs Tesla” directly. It is: Tesla vs Everyone Powered by Nvidia

@raines1220 I’m not losing any sleep about this. And I genuinely hope they succeed.

I have to admit that I feel a bit worried when I saw Nvidia release Alpamayo. It is, in my opinion, the first real competitor to FSD. First of all, it does not rely on LiDAR, which is already impressive and better than all other losers out there. Then, the concept of VLA is my exact understanding of what FSD 14.3 would be. And I believe Nvidia is doing really well at training general-purpose, small-sized VLMs. So, I don’t think it is very obvious who will train the better VLM for driving between Nvidia and Tesla. I’d guess Tesla, but you just can’t be so sure because it is not trivial to predict this part. Elon and Ashok surely both actively engaged with this news, and basically confirmed that Nvidia is doing exactly the same thing that Tesla is doing, with the only difference being how far they have gone along this technical path. I understand that going from 99% to 100% is hard, but first of all, Tesla isn’t at 100% yet, and second of all, the fact that Nvidia could reach near 99% without any fleet kind of proved that it is no longer critical to have large-scale fleet data for training and evaluation.

@Kyrannio Hey Hi, how are you

What are some words we should bring back this year?

Can consciousness exist without memory? @neiltyson

@RobinhoodApp Kind of disappointing when you do your part and this shows up

@MinChonChiSF @vllm_project is it because, it allows models to process multiple requests in parallel while guaranteeing identical outputs for each request?

@vllm_project Batch invariance is crucial for scaling LLM inference.

🚀 No More Train–Inference Mismatch! We demonstrate bitwise consistent on-policy RL with TorchTitan (training) + vLLM (inference) — the first open-source run where training and inference numerics match exactly. It only takes 3 steps: 1️⃣ Make vLLM batch-invariant (same seq → same output regardless of batching) 2️⃣ Ensure forward passes in training use identical kernels as inference 3️⃣ Add custom backward passes in PyTorch ✅ Verified on Qwen3 1.7B + GSM8K: • batch_inv_ON (bitwise exact) → KL=0.0, faster convergence, higher reward • batch_inv_OFF → reduced reward, instability We audited every op, imported vLLM’s fused kernels (SiLU MLPs, RMSNorm+residual), and wrote matching backward passes. Run is fully on-policy, deterministic, and reproducible. Next: • Unified model code • torch.compile support • Perf tuning (current bitwise RL ≈2.4× slower) • Broader model + op coverage 🔗 blog.vllm.ai/2025/11/10/bit… #vLLM #TorchTitan #RL #LLM #AIResearch

@shaneparrish without deviation from normal, progress is not normal.

You can be normal, or you can be great. You cannot be both.

@neuralink can we extend general relativity to include other forces so that we can map out the neural connections of everyone who has lived or ever will live?

@gaganbiyani what signal do you now trust most when estimating AGI timelines?

@signulll @sterlingcrispin Powered with Gemini & Vertex AI, I Was able to build a production grade financial analyst agent from Zero in less than 30 minutes! AI went from science fiction to daily life faster than we noticed

precisely. you only need to discover a small arbitrage & then scale it up like this. ppl think you need $50k right away, the reality is business does not work this way. you need to find the lowest form & scale it up. & with ai the marginal cost of personalized production dips radically.

a long time ago my favorite interview question which everyone absolutely hated me for asking was: “how would you make $1k on the internet today assuming your life depended on it (nothing illegal)?” even 5-7 years ago this was pretty difficult to do & totally non obvious to 99% of ppl. but today we’re in a dumb arbitrage era. if you can truly wield the current toolchain that includes models, automation, & just moderate levels of distribution you can print $100–500k/yr w/ shockingly little institutional work. like, i can think of 10 things i would do right now if i had to make $50k tomorrow. ridiculous times.

@elonmusk @ArtemisConsort Humans: this is technically allowed, but it’s wrong Current AI says: this matches the reward function That difference is the AGI gap.

@ArtemisConsort He lost his marbles a long time ago

If Yann can reason about intelligence in general, then he can reason about the things intelligence can do in general, which means he is a general intelligence. He says we suck at chess. And we do, compared to modern computers. But we can play chess, and moreover we can understand why certain strategies tend to succeed or fail. We can do the same for any other game. We might be better or worse at any particular game than some other entity, but we can play all games. This is not true of e.g. tigers. They aren’t just worse at chess than us. They can’t play it at all. There is very clearly a leap to universality in intelligence, as @DavidDeutschOxf describes. There are limits. Turing completeness, from which Deutsch and others analogize, only applies to systems with infinite memory and time. But there are axiom schema that, given infinite tape, are general, and ones that are not. Similarly, there will be architectures for AI that, given infinite tape, *would be* general, and ones that would not be. We may have already found the general architecture and just need to work on tools and memory. Or we might need some new insight. But to doubt the in-principle possibility of general intelligence is very bizarre.

_ in java 25 refers to 'I don't Care'