I'm surprised more people are not talking about the fact that the two best models in the world, Gemini 3 and Claude 4.5 Opus, were both trained on Google TPU clusters.
Presumably, inference can be done on TPUs, Nvidia chips, in Anthropic's case, new stuff like Trainium.
Google is a direct competitor to many LARGE buyers of GPUs and therefore a non starter from a business perspective. In addition, many companies cannot single source due to risk considerations. Hardware is different because buying GPUs is a capital investment. You own the asset and revisit the supplier only at the next refresh cycle, not continuously as with rented compute.
I despise Facebook and all that it stands for, but if the surplus value that it has extracted from humanity over the last two decades is reinvested intelligently into nuclear energy, I'm actually okay with it.
Despite the hype that you see on Twitter, the hard tech startup scene is actually incapable of large-scale engineering coordination on the level needed for a nuclear power plant, or even a gas turbine.
If any innovation on fission reactors is going to be successfully commercialized, we will need to see billions of dollars of investment over medium to long time horizons.
Of course, the millstone around the neck of nuclear power is that it's a dual-use technology. There's probably a lot more behind the scenes that's been done to stifle the industry effectively for non-proliferation reasons, but masquerading as cost, regulatory problems, environmental concerns, etc.
I would buy a Cybertruck tomorrow if it had a gas engine. I would buy a $10,000 or $15,000 gas generator add-on if it enabled unlimited range (provided I have gasoline).
There are just too many places, even in California, where I have to limit my trip because of electric range.
i actually love the way it looks. i dont own it but i find it some inspiring that a company had to have audacity to create something so futuristic - what are you driving that is so beautiful ?
Interesting. I don't find the TC to be ugly; I find it to be dissonant. It hurts my brain like missed harmonics in a musical performance do. Here are my reasons:
First thing I noticed on reveal day: it looks like a star ship from the the 1985 game, Elite.[0] It's a 3D model of a space ship for a computer that could barely keep up.[1] This design was a great starting point for a child's imagination, but even as a kid it was always assumed that this was the best we can do for now. The future would be far less disappointing. Verdict: this design isn't futuristic; it's nostalgic.
Looking down, I saw that its beautiful, shining, crystalline, space-going shuttlecraft aesthetic sits on matte, round, rubber wheels tied to the ground. It wants to fly, but it can't, and that is sad.
A few months later, I saw the unfortunate resemblance to industrial garbage receptacles usually kept out of site behind decorative enclosures. I realized that while designing one of those enclosures. Then the memes came.
I actually prefer a version I saw that was mounted on tracks for arctic environments.[2] It says, "I am a raw shard of ice carved from a massive glacier," and it pulls it off quite well.
I don't find the idea all that audacious. Several EV trucks were already in the works. The Cybertruck is unique in form, but certainly not in function. There's precedent for sloppily-made stainless steel wedge-shaped American cars [2] thought up by executives on too many drugs [1].
The execution? Whoever figured out how to get a stainless steel wedge on stilts through NHTSA testing deserves a raise. That's sorcery.
Given recent high profile redaction events, I think one simple use of AI would be to have it redact documents according to an objective standard.
That should in theory prevent overly redacted documents for political purposes.
An approach that could be rolled out today would be redacting with human review, but showing what % of redactions the AI would have done, and also showing the prompt given to the AI to perform redactions.
I don't think the commentor above is saying that an AI should necessarily apply the redaction. Rather, an AI can serve as an objective-ish way of determining what should be redacted. This seems somewhat analogous to how (non-AI) models can we used to evaluate how gerrymandered a map is
Whether or not the patent was actually granted in this case, I have not been able to think of a compelling reason to have patents for software. In fact, I think most intellectual property laws need to be seriously rethought.
If the objective is to maximize investment by protecting successful results, I don't think our system is doing a very good job.
Who can do this with good data controls? I don't want to have to dig through the fine print of some Terms of Service page to figure out if a sequencing company is going to save a copy of my genetic code for possible future use.
I sequences my genome about 10 year's ago using illumina platform for ~1200AUD. We used a university sequencing facility. They were happy to extract and sequence the dna using a shotgun approach. Depth was 5x and I think we achieved about 90% coverage. It was just for fun.
The issue with this approach is that you'll receive raw data that needs to be processed. Even after processing you'll need to do further analysis to answer your questions. After all this, I'd be suspicious of the results and seek a medical councellor to discuss and perform further tests.
I'd advise on thinking what questions you want answered. 'Sequencing your genome' sounds amazing but imo you're better off with seeking accredited tests with acrionable results.
The really difficult part of sequencing is after the Vcf. Before that is trivial, you can plug your fastq raw data in some Nextflow pipelines and in a couple of days of computing get plenty of "results" to be busy exploring for years.
> if a foreign power can buy the platforms and adjust information flows to "shift" public opinion their way.
NO.
I will never accept this premise and nor should any American. The people ARE the government and they can be influenced any which way they want to be. There shouldn't ever be any such thing as the government "protecting" people from influence. If the people want to all tune into some foreign broadcaster all day and love everything he's saying. That's how the country will be.
The second you accept "oh foreign influence is manipulating our voters to X or Y so it has to be stopped" you are signing the death warrant for free speech. This is EXACTLY the justification used in all sorts of authoritarian garbage places to suppress information and restrict speech
I agree with you on this. People who don’t believe this fundamentally do not believe in democracy.
You may worry about citizens being sufficiently educated, you may worry about them having access to enough information to make good decisions. But to restrict information for fear that they may make the “wrong” decision—the one you don’t like—there is nothing democratic about that.
Foreign voices and perspectives are critical to understanding the world as a whole and making informed decisions. If the population isn’t ready to take in that information and sort the wheat from the chaff, then they aren’t ready for democracy.
Sure, but there's a difference in intent and degree between "foreign influence is manipulating our voters" via political content on TikTok vs a targeted state sponsored campaign run by intelligence agencies.
There are some common sense guardrails around elections in a democracy, and foreign influence targeting the actual election process is generally viewed as over the line of acceptability.
> foreign influence targeting the actual election process is generally viewed as over the line of acceptability.
Nope. "Influence" is just content. Free people can consume whatever content they want, even if it's about elections. Period. End of story. Either you're free or you're controlled.
That's a severe oversimplification. The world isn't black and white.
Take free speech, which has certain restrictions in most societies. Some societies draw the lines further than others.
Likewise, when it comes to elections, the election process is controlled to some degree. One can't, for example, broadcast misinformation about mail in ballots, because that would threaten the integrity of the democratic process itself.
There are countless ad absurdum arguments that could be made here, but that one will suffice. Say a foreign entity broadcasts misleading information about mail in ballots as part of a wider campaign to convince certain sections of the population that the post office is changing votes. Would banning that activity change a society from a "free" state to a "controlled" state like some sort of binary switch? Of course not, and typing "Period. End of story" is a completely asinine and amateur argument to support that. Devolution into semantics.
What I'm highlighting is that in most democracies, there is a sacrosanct ring around the democratic process itself - arguably necessary to at least some extent - and in the context of misinformation/state meddling/etc that would fall under the category of "election interference" if a certain bar is cleared. Ideally a high bar, so the red tape is as minimal as possible, but when it comes to state intelligence agencies specifically targeting an election itself, there is not just a coherent argument for it, it's the way it is in most democracies worldwide.
You and I both know the US was not interested in forcefully taking over TikTok because of the possibility they could broadcast misinformation about mail in ballots near to election time
That is a crime because it is fraud (election fraud). If this arose on TikTok the great thing is, you can prosecute the creators who post such things. China, by promoting such stuff, in fact would make the prosecution job easier.
People in these comments talk about "influencing" people to have certain thoughts and thus change how they vote by showing them stuff that might change their minds. That is the very idea of free speech itself and it should never be under government control. Not fraud but exposure.
Presumably, inference can be done on TPUs, Nvidia chips, in Anthropic's case, new stuff like Trainium.