You need both, but it's not exactly realistic to have metro lines everywhere. The west side is pretty dense and has a lot of commercial/business activity so the value per dollar is high. By comparison the east side is super spread out and has a low(er) residential density so the value is far lower. Not to mention a good chunk of the residents on the east side aren't in favor of metro lines in their neighborhood (the stalled Whittier extension of the E line).
It's a pain for me personally since a group I'm involved with is active in the San Gabriel Valley and since I don't have a car I have to rely on buses. It's not an exaggeration when I say a trip for me that's 5 miles west would take double the amount of time east. I'm just glad the planned subway to Van Nuys and the San Fernando Valley up north seems to be going well.
A worst case scenario I feel is that the government could restrict inference providers within the US to run only approved/American LLMs, which would be a huge deal since the only recent American OSS model is Gemma. I could see OpenAI/Anthropic/Google lobbying for that though…
My thought as well. They will approve every new American trained LLM but they can't control the release of free Chinese LLMs. Therefore, the only card they can play is to simply make Chinese LLMs illegal to use for American companies and Americans.
Ultimately, this will grant more power to OpenAI, Anthropic, and Google due to regulatory capture but it hurts the AI industry overall.
Only the us AI industry. It will put the US in a ditch with it's only asset being the ability to surveill its citizens for negative money and negative innovation. The rest of the world will keep spinning just fine
Nah there's literally no reason geopolitically or economically to do that for Europe. The US has more or less entirely botched everything except it's military applications for AI. It's endangering the whole economy in the process too.
Not for the same reasons (such as the direct propaganda I expect from the Trump government), but if there is a precedent I can definitely see propositions to have some sort of regulator vet models before release to protect young people or similar
Chat control also failed. There's a lot of hope still.
The thing goes both ways. They have to secure their people from Russian and American propaganda that will be coming by the petabyte once a few more us data centers go online. The US is trying to elect fascists in Europe.
At the same time it's a terrible practice for privacy and human rights. Especially in the wrong hands.
They know very well that China is going to keep releasing world class models at 1/20th the price and 5-300x smaller in size. They also know they screwed up by going full technofacism and there's no way back because of the trillions invested in oligarchs and it endangers the entire economy.
China can’t keep doing that. This is essentially a capture the market ploy that is government back from them at this point at current DeepSeek prices they would have to make a massive loss.
I think you misunderstood what China has been doing to make the progress they have. They've invested far less and have actual applications because they are the world's largest manufacturer.
In the us we have products we sell to china to automate their factories. China soon wont need those. The US goal of laying off anyone who thinks for money is really different than chinas goals of automating product manufacturing.
Deepseek costs less because it actually costs less. Chinas electrical infrastructure is so much better than the uses. Meanwhile the us has ai data centers running on effing gas. On literal gas generators. The only budget discussions for infrastructure in the us are basically for the DHS too. It's not sustainable.
They aren't spending money like we are. They've been able to do what they have with an embargo on Nvidia cards even.
I think we will see model size shrink more and more and become more efficient. Ideally to the point where they run on high end computers and not data centers. That's the future in my opinion.
At that point you could run them on your phone or chrome book for free or with ads like Google search. Or pay for privacy
Aurareturn. I don't think they care. Geopolitically China stands to gain a lot by commoditizing it's complement. They don't need a booming insanely profitable ai industry. They are not trillions of dollars leveraged in. I think for them the applications matter more. And they can do just fine with the us scrambling for a few decades. They don't think short term the way the us does. Neither do most countries. They also don't do the individualistic burn everything down for one trillionaire thing.
Sure but those are still profitable enough without trying to take over the world. Even with free models. Not every new idea is worth 20 trillion dollars. I think most small companies are happy to take home a few billion.
Alibaba was almost measuring profit in the trillions of dollars profit last year. They make the qwen models. Sure their profit is driven by another vertical, but I don't think they are in any danger of having cash flow issues. Supposedly they have already broken even because they had serious business needs for LLMs. And again that vertical is complimented by their investment. It draws in ad revenue, customer support, etc.
I'd love to be wrong but I don't see how Chinese frontier models will keep getting released for free. I'm pretty sure Qwen no longer releases their flagship models to the public.
They may not see the economic value of releasing models for free and then hoping that you'd use their API. Once they catch up more to US models, they will inevitably go closed source. That's my prediction.
I could see a much more restrictive licensing agreement before going full closed source. It could be a scenario where hyper scalers such as AWS or Azure will gain far more value from free Chinese models than Chinese labs such as when AWS often gained more than for-profit open source software than the creators.
I’m primarily a coding instructor but I also teach lessons based around Kerbal Space Program. From that angle I’m pretty happy where these devs are going with the sandbox and level of freedom they plan on giving users. One reason my students seem to really like KSP are all the different things you can do/build (especially space planes). While it does have a high difficulty curve when it comes to the controls, once they master them they all seem to appreciate the level of control they have. From my perspective, I feel that instead of trying to make certain systems more accessible/simple they should go in the opposite direction and offer more control/options (preferably with a decent tutorial though).
However, something I really hope the developers plan on making a macOS port. I use a MacBook myself, but I’ve noticed the vast majority of my students also have Macs. I have no hard data to back this up, but I suspect a larger than average segment of users interested in this game are on Mac.
I relied on Mechjeb too much in my game, so I did a run using KRPC and built my own autopilot features with it. It was a lot of fun, and I would recommend it.
I thought it would be nice to have actual mission planning in game for things like complex maneuvers, orbital slingshots, Lagrange point orbits, and the like.
Then launch countdowns, maps, exact timed burns that you'd still have to do manually but you could seemingly do so much more.
Or maybe this kind of thing already exists with mods.
Mechjeb does a lot of automation for complex maneuvers like planetary transfers, ascent autopilot, target interception, etc..
You can even do a porkchop plot for interplanetary transfer.
but lagrange points are not possible in KSP because of the Sphere of Influence Model.
and orbital slingshots are not included, I assume because they require an intense bruteforce calculation.
Aha ok! Ambition reignited, it's not that it's trivial but it's very possible to simulate these things and there are plenty of valid shortcuts. The "BMW" astrodynamics books I learned a large chunk of it from was published in 1971. "Sphere of influence" is a lot more "things you have to do with pen and paper" than things you're forced to do with modern computers.
somewhere they have an FAQ where one of the questions is about Mac support, i remember the response being something like "most of the developers use Macs so it might happen"
I love EVA, but I’ll cautiously recommend it. I feel like there are two sides to it: the mecha/alien/monster sci-fi side which has an amazing aesthetic, and the personal drama focusing on self loathing and loneliness. I think the first side is the most attractive to most people, but what really sticks with me to this day is the latter.
If you do end up watching I have to warn about the watch order. There are two timelines, the original TV series plus the movie “End of Evangelion”, and then the “Rebuild of Eva” movie series which started as a complete reboot but somehow ended as the ultimate Reboot/Remake/Sequel to the original stuff.
I understand, I had started watching these evangelion summaries thinking oh cool robots but then, I have been aware of the loneliness and other things within the series which were thought-provoking but also not really at the same time. I
I don't watch much anime (shocking within my generation which seems to be anime-driven?) but ironically I have also watched berserk's golden age arc also because of an edit that I had watched before all the horrificness is revealed but that's the reason why I haven't really read more about berserk other than the golden age.
Yu-Yu-Hakusho, death note, AOT and bersek/(evangelion? Only watched documentaries explaining it) are the only series that I have watched/listened about a lot in this sense I suppose.
With berserk and evangelion, perhaps there is something about the style of both of these, both lure you in with a particular sense of art but both have some aspects of darkness hidden beneath the pages.
There is something about first discovering it even knowing through comments that things would turn south but I still decided to watch berserk because it so captured my attention. I wished for it to be a bland piece where things turn happy ever happily but I think that berserk's decision to not do is somewhat fascinating too. maybe the lure to watch these increases knowing the fact that these aren't just single dimensional things but rather multi dimensional with deeper topics.
Evangelion with concepts like loneliness and I would associate berserk with pain and hope (hope to still fight even after so much pain)
Edit: should probably mention Its late night and I should probably sleep and I didn't really intend to send this message, I sometimes write somethings and don't send them (but save them for myself to read in documents)* because I find that I think deeper about something during this and I accidentally pressed enter, thanks for reading this but I am sure that I might not be able to give my viewpoint away but I am grateful that you read till the end. Looks like the world wanted me to send the message so I did :-D
Both are great shows made by GAINAX, but I highly watching suggest their prototypes as well: Gunbuster and Diebuster. All four are great Mecha shows made by the same studio and it's fun spotting the similarities across all of them.
What I consider to be a spiritual successor to these GAINAX mecha shows was the most recent Gundam series "Gquuuuuux" which shared many staff members from them and has plenty of homages that were fun to spot! Also had the same mechanical designer as the Evangelions so I got a kick out of that.
For my customers I use regexes to block them from potentially publishing personal emails/phone numbers to their websites but I really wouldn't mind running this in addition just for the extra peace of mind. I don't have a GPU on our server, but I hope this is light enough of a model to handle CPU only inference on less than 2k tokens at a time.
I use the Throne bathrooms all the time here in LA. They’re really conveniently placed next to a bunch of Metro stops and are pretty frequently used. I’d actually argue that for busier stations (North Hollywood) more should be added!
The way it works is that you text a phone number (facilitated by a QR code) which Throne you want to use and it automatically opens. These bathrooms play classical music and are touchless so you don’t have to touch any knobs or buttons, you just wave over a sensor to operate the sink or open the door. It’s overall super frictionless/convenient and I agree with the article on how well they’ve been treated (I haven’t seen any graffiti or vandalism yet) despite their frequent use.
35B (35 billion) is the number of parameters this model has. Its a Mixture of Experts model (MoE) so A3B means that 3B parameters are Active at any moment.
As someone who spins up docker containers where I use the Anthropic Agentic SDK to build Jekyll websites for customers, I don’t see much of an appeal. I didn’t find it that difficult to set up the infrastructure, the hard part was getting the agents to do exactly what I wanted. Besides, eventually I might want to transition away to another provider (or even self hosting) so I’d prefer having that freedom.
It's a pain for me personally since a group I'm involved with is active in the San Gabriel Valley and since I don't have a car I have to rely on buses. It's not an exaggeration when I say a trip for me that's 5 miles west would take double the amount of time east. I'm just glad the planned subway to Van Nuys and the San Fernando Valley up north seems to be going well.
reply