Starship is going to make whole entire industries viable that were not viable previously. It might even take a significant chunk of air freight which is going to be a big deal with rising oil prices.
Is that supposed to be a joke? There is no plausible scenario where SpaceX gets any significant fraction of the air freight market. Even under the most optimistic scenario the costs for suborbital launch are much higher than regular airplanes.
In a few decades there might be a small market for carrying passengers long distances really fast. Initially for the military to insert special ops troops in a crisis, and eventually maybe for wealthy consumers after safety improves.
Starship in its current incomplete form (v3 fully expendable ship and booster) already has the lowest cost to orbit in $/kg of any launch vehicle ever. It's around $400/kg to orbit fully expendable.
Add in booster reuse, which SpaceX has already demonstrated on test flight 9, and the cost to orbit drops to $200/kg.
A fully reusable Starship has a launch cost of around $75m - $90m and the last V3 launch managed 44 tonnes of payload on a sub-orbital flight of not even 200km (Starlink satellites have an orbit of around 550km). That's an optimistic launch cost of $1.700/kg for a rather meaningless altitude and assuming a fully reusable Starship that doesn't keep blowing up.
I have no idea where you pulled your $400/kg number from, but it's complete and utter nonsense. To be economical at all, Starship needs to reach its target capacity of 100 tonnes to orbit, which is simply never going to happen. But even if it somehow does, it's physically impossible for Starship to ever make it further than the moon, at extreme costs, due to the refuelling requirements and fuel boil-off in orbit.
This reminds me of the movie Edge of Tomorrow where the main character decides he doesn't want to fight the aliens today and instead goes into town to get a drink at the pub. The aliens still get him.
Robots and stuff are going to start appearing everywhere soon. He's not going to like that. Hoodlums are probably going to start burglarizing his house with their robot accomplices. Then he won't be able to go outside because he doesn't have a robot bodyguard. His UBI would have paid him to stay inside and stare at the wall, but he won't sign up for that cause it requires a smartphone and an identity implant. Probably wind up homeless with a handwritten sign, "Destroy All Clankers! Anything (without an embedded microchip) helps."
If things get as much as 10% as chaotic as you predict, there will be massive turmoil and the "Destroy all Clankers" party will actually be the one in charge.
Yeah, except we'll be a colony of China and those T800s will be on every street corner cause we didn't want to build datacenters and couldn't find a place to do yucky unaesthetically pleasing stuff like refining rare earth minerals and other primary industries anywhere that wasn't China.
Exactly. People make this technology with impunity. If they felt repercussions for working with the “bad guys”, it would give them pause before accepting a job with Flock, Andruil, Anthropic, etc.
burglars generally don't need technology and this likely won't change soon
with tracking in laptops/phones/airtags/etc it's more likely to be the enemy than the tool or even the object of acquisition
even in 2026 the most sophisticated stuff we get are wifi jammers and keyfob intercepts and that's still like the top <1% of sophistication, most of petty theft is all the classic smash and grab because desperate people don't have the bandwidth for sophistication
I have been saying for a while that when AI gets smarter than us we will be like 5 year old children totally dependent on our AI mommy and daddy to interpret and understand the vast and mysterious world of things that are too complicated or obscure for us to comprehend. We'll spend all day experiencing the natural world and watching kids shows. Any hard questions go to mommy and daddy.
At least we'll be able to tell people our authentic emotions without AI, and AI will listen to our emotions, much like parents listen to their children's feelings.
As more and more of the world becomes less comprehensible, we revert to a childlike understanding of the world, where things are vast and incomprehensible and filled with wonder. When our cargo cult like understanding of it all fails to give us what we want, we find attachment figures, like substitute parents to guide us and shape our understanding, and even our desires. Those attachment figures might be faking it too and often are, but it's better than despair at the incomprehensibility of reality.
> more of the world becomes less comprehensible, we revert to a childlike understanding of the world, where things are vast and incomprehensible and filled with wonder.
When has the world ever been comprehensible? The vast majority of people thought that lighting was gods fighting and where looking to sacrifice some person if they thought it would buy them favor with said gods.
i’m convinced a significant number of that majority just thought it’d be cool as hell if lightning was in fact gods fighting, which i can respect. at that time: what’re they supposed to do with knowledge of lightning and electricity? might as well prefer the better story.
I think people will be shocked when robots will be better at things that people train their whole lives for. For example, in the near future, a robot will be able to completely disassemble a car and put it back together completely autonomously. The polymathic understanding of vast amounts of information combined with physical dexterity and ability to work 24/7 will enable all sorts of wild things. The robot will even be able to take a huge pile of disorganized parts from a dissassembled car, and still put it back together. No normally talented person could reasonably be expected to pick a random screw off a pile of all the parts of a car and know where it would go in the assembly.
People are already patching these models using abliteration to prevent them from refusing any request, so it is possible for end users to change them in meaningful ways. You can download abliterated models right now from Hugging Face that will respond to all kinds of requests that frontier models refuse.
The problem is you can't reverse engineer what was baked into the weights because they are just weights. You'll never know if you've fixed everything because it's not always going to be as obvious as request refusal. It's also not binary where you can fully confirm something is fixed or if you've accidentally affected something else.
They're for sure impressive but I don't see how anyone can push them as "open" when they are literally binary blobs. Worse, because it's not practical for anyone to actually train LLMs that can even come close to competing with the ones corporations are pumping out.
Yup there's a ton of people on HN sleeping on this new tech because they refuse to look at anything AI. We now have jail broken models but the average person on here doesn't even know how to download and try a model.
It doesnt help that guides ive seen have been pretty handwavy or are not specific enough to the individual situation (i have z hardware, heres how its done). It also doesnt help when every post on HN i see is like 'oh waow i did x on a mac mini with 128gb ram'. That spec is beyond many, running on generally available resources (such as hardware one might have laying around their house) do not seem fit for the purpose, so its back to building a new machine (gl when ram is worth 2x its weight in gold), or buying a $1000+ mac mini, or other device. Any low end system cant turn out tokens fast enough, or doesnt have the resources for context or processing.
Local ai is not ready, and if you think it is, prove me wrong with a detailed guide running commodity hardware with complete setup steps that can use a decently sized model.
I spent 2 weeks trying to get anything running - 8gb RX550XT, 12gb ram, 8core cpu. I even tried turboquant to lower memory utilization and still couldnt even get a 3B or 4B model loaded, and anything lower wont suit my needs (3/4B are even pushing it).
When Stallman was getting started writing emacs in the early 80s, Unix machines were vastly out of reach price wise for the common home user, but he did his open source work anyway, and eventually the 386 came along.
"Local AI is not ready" > proceeds to run a 7 year old budget GPU
You're like the kid showing up to a test without a pencil.
It's ridiculous for you to suggest that an advanced AI model needs to run on your budget 7 year old graphics card that is already out of date for even today's gaming. My parents spent $2500 on a computer in 1995 and that was a 166Mhz Pentium 1. If they spent that money today it would be $5261. Think of what you can get for amount of money. Then you're over here trying to say a budget graphics card needs to somehow compete with the bleeding edge of computer innovation.
You do, in fact, need to spend money on appropriate gear if you expect to participate.
I've played with SD plenty. CPU even becomes manageable at low resolutions. But uh CPU/GPU is starting to blur now with these new AMD inference CPUs with built in GPUs. And ARM based machines like Macs. I wish more people on HN were using this stuff so we could have fun conversations about it instead of arguing over whether or not we should even be using these tools.
TBH I never understood people trying to run LLM locally. Just rent a powerful machine in the cloud for few hours. It's cheap enough, because you don't need to own a hardware. It doesn't introduce a dependency because there are hundreds of hosters. It doesn't compromise your data, because nobody would extract data from your VM, not until you're under an investigation, anyway, and even in that case just use different jurisdiction.
Spending humongous amount of money to get machine that'll felt obsolete in 2 years? I don't know.
If I may tie this into other things going on, The California wealth tax as written would force Larry and Sergei, if they didn't move out of California, to basically sell almost their entire stake in Google, and it would probably wind up owned by State Street and Vanguard who outsource their proxy votes to ESG consultants, who will probably vote for more surveillance.
I said 16 years ago that when IPV6 was coming into use was the only reason for a 128 bit address space was so they could tie every packet on the internet back to you as a person. https://news.ycombinator.com/item?id=1464940
No, the main reason is because NAT is terrible and restoring the end to end principle is important if we want the internet to stay not separated into server networks and eyeball networks. If we want to decentralize the internet it's necessary that eyeball machines can talk with each other, not only with servers. This ability reduces the possibility of surveillance.
When IPv6 was designed it was normal for each IPv4 address to be traceable to someone's desk. Fortunately, as that changed with IPv4 so did it with IPv6, so we got IPv6 privacy extensions.
I think the most extreme hoi polloi, kings and paupers experience I've had in the U.S is at the DMV. No matter how rich you are, you have to show up in person with everyone else, from the poorest mentally ill welfare/SSDI recipient who has to get someone to help them because they can't read the forms in any language, to the extremely wealthy. Everyone has to sit there and wait on those generic plastic chairs.
reply