We should already be at that point. We have already seen LLMs' potential to inadvertently backdoor your code and to inadvertently help you violate copyright law (I guess we do need to wait to see what the courts rule, but I'll be rooting for the open-source authors).
If you use LLMs in your professional work, you're crazy. I would never be comfortably opening myself up to the legal and security liabilities of AI tools.
Wow, what a dishearteningly predictable attack.
I have studied computer architecture and hardware security at the graduate level—though I am far from an expert. That said, any student in the classroom could have laid out the theoretical weaknesses in a "data memory-dependent prefetcher".
My gut says (based on my own experience having a conversation like this) the engineers knew there was a "information leak" but management did not take it seriously. It's hard to convince someone without a cryptographic background why you need to {redesign/add a workaround/use a lower performance design} because of "leaks". If you can't demonstrate an attack they will assume the issue isn't exploitable.
Two Vonnegut novels—God Bless you Mr. Rosewater and Player Piano—fundamentally shifted the way I view the world.
The novels primarily discuss the economy, automation, and human wellfare. When I was young I defaulted to a laissez-faire economic mindset, and basically assumed automation and technology would always make our quality of lives improve. I was very much in the Ayn Rand club on economic and moral issues. These books were ultimately what made me reflect and consider the other "spiritual" (in the sense Vonnegut uses the term) aspects of human wellfare. Vonnegut was my introduction to humanist thought, and I owe the vast majority of my personal moral development to the influence of these two books.
Ah, yes. The famously singular "westerners" who all 100% agreed with every foreign affairs policy of their government over the past century.
... on AMD's most powerful GPU.
I mean... At the current state of the game, 0% of gamers will be playing at 4K/High settings.
CSGO cases pulled $1 billion revenue in 2023. The steam store brought in $8.5 billion in that same year. That's a 30% cut of all sales traffic on steam vs. in-game loot crates on a single title.
Loot boxes pull insane numbers. And yes they exploit children and problem gamblers. Love to see so many Valve fans downvote you :/
At least there is a big (ish?) player in the Chromium-sphere pushing back against this.
The more browsers that don't initially support this, the slower adoption by web sites will be. If enough of the browser market share remains incompatibe, and if we're lucky, maybe this technology won't stick.
That's significantly worse privacy-wise, since Google gets a copy of everything.
A recovery email in this case was used to uncover the identity of the account-holder. Unless you're using proton mail anonymously (if you're replacing your personal gmail, then probably not) then you don't need to consider the recover email as a weakness.
I'll just add that "designed to be burned up" is the correct approach to these types of satellite constellations. SpaceX has that aspect correct, at least.
Agree with everything else. Musk is a batshit egomaniac, and letting him dictate use of large infrastructure is careless. Government subsidies should entail a certain public influence over the operation of the system.
That is one theory, based on a conversation captured on another (not-directly-involved, but on-site) office's bodycam footage. It isn't really conclusive, it's on-scene hearsay from what is likely the downstream end of a game of telephone.
The more productive avenue for discussions, in my opinion, is to consider whether firing pepper balls at non-violent individuals is perhaps negligent or reckless use of force, that escalates the situation without solving anything.
Her manner of speaking reminds me of the sermons you get at 'modern'/nondenominational churches here in the south. Just the way phrases are timed, the intonation, the need to make every minute factual statement sound emotionally profound...
I have to wonder if she is consciously trying to speak in that way. I don't know why they would think that was a good approach for a political speech lmao. It's just so bizarre I can't actually process it.
I read "fully half" as actually >50%, not rounding up
SBF's peak was a few years ago. This year all he's done is show during his trial how deluded these techbros/EAs can actually be. At least SBF had the common courtesy to remove himself from public life within 5 years. Style points for the life-in-prison ending, while simultaneously killing mainstream crypto.
We're stuck with Altman for the foreseeable future, and now with a recently purged OpenAI board that will let him continue the industry-wide commercialization of copyright infringement (but only of the laypersons' IP. Better not ask it to draw Mickey Mouse, though).
It's also really unclear where OpenAI lies on the EA/Longtermist/E Acc pipeline. Altman is likely letting whackos have some pretty serious power.
The reason the article compares to commercial flights is your everyday reader knows planes' emissions are large. It's a reference point so people can weight the ecological tradeoff.
"I can emit this much by either (1) operating the global airline network, or (2) running cloud/LLMs." It's a good way to visualize the cost of cloud systems without just citing tons-of-CO2/yr.
Downplaying that by insisting we look at the transportation industry as a whole doesn't strike you as... a little silly? We know transport is expensive; It is moving tons of mass over hundreds of miles. The fact computer systems even get close is an indication of the sheer scale of energy being poured into them.
The example is pretty standard, but I feel obligated to caution people about the author (just because he's linked to here and some unassuming people might dive in).
Scott Alexander falls loosely under the TESCREAL umbrella of ideologies. Even in this article, he ends up concluding the only way out is to build a superintelligent AI to govern us... which is like the least productive, if not counterproductive, approach to solving the problem. He's just another technoptimist shunting problems onto future technologies that may or may not exist.
So, yeah, if anyone decides they want to read more of his stuff, make sure to go in informed / having read critiques of TESCREALism.
It's especially infuriating when you have a giant like Microsoft rolling Electron on their flagship applications (Teams), and then deprecating support for some platforms (Linux). What's the point of your nice, memory-gobbling, platform-agnostic app framework if you're not even going to use it to provide cross platform support?
Lots of weird polish issues in my opinion... One that really peeved me was (for a while at least) you could search for a message, but there was no way to jump to that message from the search results. So you couldn't read the context unless you scroll all the way back up.
But primarily it's that the mechanics are different from things like Slack and Discord in ways that are just less intuitive.
Channels function more like announcements + comments rather than a chat—you are really shoehorned into posting a "Topic" and discussing it in the replies. There's no way to carry a linear conversation in a channel otherwise. And to load replies you have to keep clicking "see more" as if this is a social media site, so it's very annoying when your 800+ comment critical discussion happens there. Not to mention notification settings aren't granular enough, so you either get hammered by all activity, or remain oblivious to discussions which may have popped up in an older Post.
What tends to happen in my experience is small working groups spawn off a group chats because the flow is better for daily conversation there than in Channels. Which, of course hides this activity from anyone not in the chat. And group chat's are entirely linear in Teams—you don't have threads the way you do in Slack, so chat history tends to get messy quick.
The channel-then-thread organization Slack uses is much more natural for the teams I tend to work on, because you just have the one main discussion which can be segmented into threads as needed.
330 micrograms per gram
That seems like... a lot. Way more than I expected or am comfortable thinking about.
Reuters mentions the execution used a facial mask connected to a tank of nitrogen.
So yeah... No CO2 removal and probably not a perfect seal against outside atmosphere. This is a pretty half-assed implementation.
It bugs me how easily people (a) trust the accuracy of the output of ChatGPT, (b) feel like it's somehow safe to use output in commercial applications or to place output under their own license, as if the open issues of copyright aren't a ten-ton liability hanging over their head, and (c) feed sensitive data into ChatGPT, as if OpenAI isn't going to log that interaction and train future models on it.
I have played around a bit, but I simply am not carefree/careless or am too uptight (pick your interpretation) to use it for anything serious.
404media is doing excellent work on tracking the non-consentual porn market and technology. Unfortunately, you don't really see the larger, more mainstream outlets giving it the same attention beyond its effect on Taylor Swift.
I get that there are better choices now, but let's not pretend like a straw you blow into is the technological stopping point for limb-free computer control (sorry if that's not actually the best option, it's just the one I'm familiar with). There are plenty of things to trash talk Neuralink about without pretending this technology (or it's future form) is meritless.
"Platform Most Favored Nation". It's a type of clause in platform/marketplace agreements that prohibit a seller from listing their product for a lower price on a different sales platform. Specifically, it prevents selling on a different marketplace with lower fees (e.g. Epic Games or a publishers own website) and passing the difference as savings to the consumer.
It's not really stupid at all. See the matrix code example from this article: https://spectrum.ieee.org/ai-code-generation-ownership
You can't really know when the genAI is synthesizing from thousands of inputs or just outright reciting copyrighted code. Not kosher if it's the latter.
Feel the same way. My Camry is a 2013—recent enough to have a simple display and Bluetooth, but old enough to predate the 'modern' infotainment systems.
Believe me, I plan to drive this car until the scrapyards run out of part donors.
Sticking two E2EE tunnels together with a plaintext middleman doesn't result in a single E2EE tunnel.
The reason the distinction is important is because the security profile is vastly different—a compromised server leads to a compromised message—which isn't true for actual E2EE services like a pure Matrix link.
Side note: the first thing you should ask of a "end-to-end encrypted" product to you is "which 'ends' do you mean?" I've seen TLS advertised as E2EE before.
ABSOLUTELY. I just recently capped off the Diff Eq, Signals, and Controls courses for my undergrad, and truly by the end you feel like a wizard. It's crazy how much problem-solving/system modeling power there is in such a (relatively) simple, easy to apply, and beautifully elegant mathematical tool.
There is an episode of Tech Won't Save Us (2024-01-25) discussing how weird the podcasting play was for Spotify. There is essentially no way to monetize podcasts at scale, primarily because podcasts do not have the same degree of platform look-in as other media types.
Spotify spent the $100 million (or whatever the number was) to get Rogan exclusive, but for essentially every other podcast you can find a free RSS feed with skippable ads. Also their podcast player just outright sucks :/
Steam has a large userbase, which offers a lot of consumer inertia to prefer games on Steam. They also have a policy where game pricing on other platforms cannot undercut Steam.
The main complaint is that this pricing policy coupled with the consumer inertia makes it difficult for other gaming marketplaces to enter the market. You cannot undercut steam unless a publisher wants to not put their game on Steam at all (which would be suicide for anything but the largest titles), so you have to sell at Steam's price point. Few platforms could match Steams' established workshop, multiplayer, streaming, and social services; all of which benefit from costs at scale and the established user content.
Imagine trying to convince a user: "Buy your game here instead. It will cost the same as on Steam. No, you won't have access to the existing Workshop. No, you won't have in-platform multiplayer with your Steam friends." Even if you had feature parity, people would prefer Steam since that's where their existing games and friends are.
Right concept, except you're off in scale. A MULT instruction would exist in both RISC and CISC processors.
The big difference is that CISC tries to provide instructions to perform much more sophisticated subroutines. This video is a fun look at some of the most absurd ones, to give you an idea.
Having express self-checkoit is great. The Kroger near me went full-self-checkout. They have large kiosks that mimmic the traditional checkout belt kiosks, except the customer scans at the head of the belt and the items move into the bagging area.
If you have a full cart, you scan all the items, checkout, walk to the end of the belt, and bag all of your items. Takes twice as long as bagging while a cashier scans (for solo shoppers), and because of the automatic belt the next customer cannot start scanning until you finish bagging, or their items will join the pile of your items.
It effectively destroys all parallelism is the process (bagging while scanning, customers pre-loading their items with a divider while the prior customer is still being serviced), and with zero human operated checkouts running you get no choice
Spin up c/notquitetheonion?
Well the general principle is that you can't be punished for behavior that was legal when you did it. Otherwise you open the door to "doing X is illegal now" and then locking everyone who was documented doing X in the last several years.
Which maybe sounds nice when it's destroying the climate... but it's less nice when it's gay marriage, alcohol consumption, owning X book, etc.
Note that the main argument Wolfire is making is that game marketplaces (buy/download the game) and game platforms (online features, mod distribution, social pages) need to be decoupled. By integrating the two, Steam is vertically integrating, amortizing the cost, and then forcing every other marketplace to bear the cost of a platform in their pricing.
If you bought a game and paid for platform services separately, then competition can better exist for both of those roles. Which is good for consumers.
I was under the impression that the policy required a game's price to be the same on all marketplaces, even if it's not a steam key being purchased. I.e. a $60 game on steam must sell for $60 off-platform, including on the publisher's own launcher.
I just went to double check my interpretation, but the case brief by Mason LLP's site doesn't really specify.
If it only applies to steam keys, as you say, then I agree they don't really have a case since it's Steam that must supply distribution and other services.
But, if the policy applies to independent marketplaces, then it should be obvious that it is anticompetitive. The price on every platform is driven up to compensate for Steam's 30% fees, even if that particular platform doesn't attempt to provide services equivalent to Steam.
concepts embedded in them
internal model
You used both phrases in this thread, but those are two very different things. It's a stretch to say this research supports the latter.
Yes, LLMs are still next-token generators. That is a descriptive statement about how they operate. They just have embedded knowledge that allows them to generate sometimes meaningful text.
I think the idea here is to bake it into construction that would happen anyway. If you just need energy storage, keep using batteries. But if you're pouring a foundation already, why not also turn that foundation into a battery?
I don't fault your interpretation. There is a reason Vonnegut uses the term "spiritual" throughout the book. At least for me, I would describe my understanding of the book to have required a spiritual/moral shift before I could really understand the image being painted.
I also read God Bless You Mr. Rosewater first of the two, so maybe that colored how I interpreted Player Piano. It is a more direct argument that humans need to be cared for, independent of their economic utility.
So when I read Player Piano, it didn't strike me as an argument against automation (which, being an engineer myself, I am entirely for), but moreso as a warning that freedom from labor doesn't alone make a perfect life. Especially in the mid-20th century context Vonnegut was writing in, it's an argument against the "American" style of automation, wherein you displace people from their jobs and discard them entirely. They serve no further purpose to your economy, and since your society is tightly adjoined to the economy, they serve to purpose to society...
So it's not really a book about automation, if if I said that in my first post. It's a book about failings of American culture, which happens to be revealed through automation. It's about the inconsistency of a society where one's usefulness to others is determined solely by their labor, and where that labor is constantly sought to be devalued and eliminated, and what the end of that process looks like for humans who want to find meaning in their activities.
I'm glad to see for once the fines are proportional to revenue, and not a fixed amount. 6% hurts.