Mati Roy

✨Important update: 312 neuroscientists were surveyed, and they assigned substantial probability to long-term memories only depending on static brain structure preservable by modern brain preservation techniques (!!!) 1/11

.@dwarkesh_sp: By 2030, it will be less expensive to monitor every single nook and cranny in America than it is to remodel the White House. “Mass surveillance is, at least in certain forms, already legal. It has just been impractical to enforce so far. Under current law, you have no Fourth Amendment protection against any data you share with a third party. That includes your bank, your ISP, your phone carrier, and your email provider. The government reserves the right to purchase and read this data in bulk without a warrant. What’s been missing is the ability to actually do anything with all of this data — no agency has the manpower to monitor every single camera, read every single message, and cross-reference every single transaction. However, that bottleneck goes away with AI. There are 100 million CCTV cameras in America. You can get pretty good open source multimodal models for 10 cents per million input tokens. So if you process a frame every ten seconds, and each frame is 1,000 tokens, then for 30 billion dollars, you can process every single camera in America. And remember that a given level of AI ability gets 10x cheaper every single year - so a year from now it’ll cost 3 billion, and then a year after 300 million, and by 2030, it’ll be less expensive to monitor every single nook and cranny in this country than it is to remodel the White House. Once the technical capacity for mass surveillance and political suppression exists, the only thing standing between us and an authoritarian state is the political expectation that this is not something we do here.”

@InterstellarUAP it doesn't need to be observed just to be entangled, and entanglement happens astronomically more often than observation @grok

🚨 Simulation Theory: The Double Slit Experiment proves particles act like waves until observed then they snap into particles. What if our reality only "renders" when we're looking, just like a video game optimizing resources? Check out this episode from The Why Files breaking it down, tying it to Simulation Theory. Are we in a sim? This could be the key to unlocking the true nature of existence! The Why Files video did a great job on explaining the Double Slit Experiment & Simulation Theory What do YOU think—real or rendered? Drop your thoughts below!

In reply to someone saying cryonics is just preserving a "copy" of you and not really you, LW/green_leaf wrote: > The argument we cease to exist every time we go to sleep also can't be disproved, so I wouldn't personally lose much sleep over that. (source in reply) 😂

@Degen_Hardy 😂

@matiroy oh, ty for the info

BREAKING: The latest data point on the log-straight-line trend still lies on a straight line

@cgtwts Google Map and Reality will converge over time

Google Maps just set Apple Maps back another 5 years.

@googlemaps i can finally retire my custom-made Gmap toilet list 😯

Finally, you can ask Maps to “Find me a public toilet nearby where I don’t need to wait in line to buy something." Welcome to the future.

@Degen_Hardy which group are you in? 😄 assuming former main orgs i'd suggest looking into: US: alcor.org sparksbrain.org (i'm on their board; it's a non-profit) Europe: tomorrow.bio you can get a life insurance policy to pay for the service

@matiroy Okay, so what do we do now? 😁

to everyone who's not sign up for biostasis → remember it might work to everyone who is sign up → remember it might not work (in its current form)

@BullMooseMemes @kaimicahmills that's not an argument but rather a reflection of a strong evolutionary prior against change (just pointing it out as seems valuable to notice it for what it is)

@kaimicahmills I will not eat the man-made meat-sack meat. That’s an abomination, it spits in the face of all that is good on this earth. Man-made horror beyond human comprehension. Total dystopian nightmare.

the ultimate solution is through technology we engineer what has been called a bodyoid: brainless animal bodies that provide as much meat as we desire without harming any sentient beings this would transform medicine - the same platform would allow us to grow organs on demand, eliminate transplant waiting lists, and produce perfectly matched tissues for each patient experimental therapies could be tested on full biological systems without involving conscious animals, regenerative medicine would accelerate as entire replacement tissues become manufacturable in the same way that agriculture turned food from a scarce resource into an abundant one, engineered bodyoids would turn biological material into infrastructure - meat without slaughter, organs without donors, and medical research without sentient suffering

love it! reminds me when @ossia was livestreaming himself coding @freeCodeCamp which is a platform which teaches coding

the human soul

@Degen_Hardy @gork

@matiroy Grok, what does he mean?

we'll fix it in post... post-AGI that is! 😄

@Degen_Hardy @grok

@iScienceLuvr unicorn from ~day 1; crazy!

Whoa! Yann LeCun's world model startup has raised $1.03 billion at $3.5B pre-money valuation

uploading a file is making it from local to a server; it's a qualitative change on how it's store, so I think the human analogy makes sense given you also change how the information is stored from bioware to hardware once uploaded then you can make copies, or maybe you can even make copies of the full human brain by "3D printing" other identical human brains anyway, that's just terminology, but that's my opinion on it

Stop calling it an "upload". People don't understand what upload means in computing and in the long chain of future technology this is going to encourage mass suicide unless it's made clear. This is a copy, nothing has been actually moved. When you "move" data in a computing system, it's copied to the new location and then deleted from the original. Fundamentally, there is no way to avoid teletransportation paradox death in computing. We should never allow an "upload" of human beings that is destructive to the original, if we develop this technology for human brains, it would be murder. There's a real risk here that something like a social norm is established that encourages upload-based euthanasia for healthy people and we need to avoid that.

@DanTurnerEvans 💡

AlphaFold won a Nobel Prize and is being used to design new drugs. How do we get more AI models like it? My new piece introduces "Natural Law Models" (NLMs): AI models trained on scientific data to uncover the rules governing nature. I also lay out what policymakers and funders need to invest in to make more of them. NLMs are increasingly essential to scientific workflows. State-of-the-art datasets are often too complex for traditional analytic methods to parse; NLMs enable researchers to extract structure that would otherwise stay hidden. They can also massively speed up simulations by learning the dynamics of known natural laws. Building NLMs requires three key ingredients: 1. Large, well-curated datasets or physics-based simulations 2. Experts fluent in both ML and the scientific domain of interest 3. Dedicated compute The challenge: existing models have already seized the low-hanging fruit. Developing NLMs for new fields requires overcoming bottlenecks in measurement, data collection, and algorithm development. While some argue that general-purpose LLMs will eventually outperform domain-specific models, the datasets needed for NLMs will still be essential — for training new models, fine-tuning existing ones, or feeding into LLMs for analysis. The investment pays off either way. Studying NLMs themselves can reveal new science. Once a model learns representations of natural laws, scientists can extract them — as equations, visual features, etc. — and add these insights to the scientific corpus. NLMs do pose dual-use risks. The same capabilities that accelerate beneficial research could also lower barriers for bad actors. Funders and builders need to take model and data governance seriously. Read the full piece here: ifp.org/nlm

@DanTurnerEvans I'd be curious to know how likely you think this brain preservation technique is to preserve enough information for theoretically being uploadable brainpreservation.org/tech-prize/

While I’m posting about systems neuro, let me also offer some thoughts on this excellent post about building brains on a computer: asimov.press/p/brains I think this piece misses two BIG obstacles 1. Animal behavior is still mostly a mystery to us. 2. Individual neurons are capable of pretty complicated computations all on their own, and their electrical properties can change in response to neuropeptides or neuromodulators. We can’t ask animals what they’re thinking, and most of our techniques for measuring neural activity happen in very contrived artificial environments. Brains evolved to help animals live in the real world, so most neurons might be effectively offline when measured in a lab animal. I suspect that we still have a long way to go to characterize the complex electrical properties of individual neurons, and that these properties will be essential for simulating neural activity from connectomes. Neurons in the brain are very different from neurons in ML models. You can fit parameters to model neurons, but model neurons for the fly, for example, still lack detailed information about the properties of receptors, ion channels, etc.