>After researchers found security holes in the devices, they quietly disappeared.
Hardcore! Now that's dystopic. Maybe I should have posted this anonymously.
2191 publicly visible posts • joined 24 Nov 2007
We don't really know what "intelligence" is, but, ultimately, there are only two possibilities. Either it is nothing more than emergent behavior from big statistics with a lot of recursion, or it is more than that.
In the first case, we might get AGI comparatively soon. It might take some new development, such as embodied AI, or it might take finding just the right architecture. But the path would be roughly correct and ought to eventually get there.
In the second case, we are not getting AGI any time soon, and all the effort spent on LLMs can only run in circles while getting nowhere. If we're very lucky, we might learn why it's not the answer.
Both possibilities are philosophically interesting, and I'm looking forward to the outcome.
Why all the downvotes to mr. 42...39?
The notion that there's no reason to expect a LLM to be good at chess doesn't seem that controversial to me. It seems a reasonable claim to me, given that chess is not a language task. LLMs are notoriously bad at math too. But even if you disagree, it's not a stupid or offensive concept.
Is it the concept that general knowledge and basic understanding of the rules are not sufficient for "a great solution" to chess? Again, is that controversial? I have general knowledge and I do know the rules of chess, but I totally suck at the game. Not as much as ChatGPT, apparently, but I'm really far from "great". There's a lot more to good chess than knowing the rules.
The suggestion that you would get a better chess program if you trained a machine learning model on chess game records, also does not sound controversial to me. I mean, of course you would. How much better, that's hard to say without giving it a try, but it would definitely be a lot better than a general-purpose language model. Actually, isn't that how the grandmaster-beating program was made?
Am I missing some subtext here?
>SAP offers concerned customers "complete sovereignty from the top to the bottom," he said.
Right. So, let's assume my data is physically hosted inside Europe, but by a US company. If I understand correctly, this is the situation he's suggesting.
Now, let's assume that I, as a EU company and/or citizen who does not particularly want to be subject to US laws and possibly does not even operate in the US, do not want my data to be handed over to the US government just because they asked.
Finally, let's assume the US company gets a Cloud Act order to hand over my data to the US government.
What exactly, dear SAP CEO, am I supposed to do in this scenario?
I'll tell you: first, hope really hard that my goverment and the US government have an agreement on what happens, and that said agreement involves them getting permission from my government. If they don't, then my data is gone.
Next, I have to hope really hard that the US government is actually respecting that agreement. If they don't, then my data is gone - but I might get redress in court, after a lot of time and a whole lot of money. But if the US government really feels like it, they'll find an excuse and I won't even get that. In fact, I might not even know that my data has been copied until long after the fact.
And when I do know, and raise hell, what's going to happen anyway? The damage is done. And I can't even drop them and go to an EU cloud provider - because we've decided we don't need them!
This is not what "complete sovereignty" looks like.
Sort of. It depends. I've lost a lot of paper stuff while moving, and a lot more is extremely hard to access because I wouldn't know whether it's in my parents' attic or in my attic, and wouldn't know in which box it is, and it's all unsorted. On the other hand, I still have a whole lot of digital documents from when I was a kid, they are actually accessible, and it's all backed up to an encrypted could storage. You could probably hack or disconnect me if you tried hard enough, but if it was something running an airport (instead of old D&D campaigns) it would be air-gapped and offline and have UPS and a gennie.
Arguably, the objection to that would be that I was clearly not storing my paper properly while at the same time being decent at storing my bits and bytes, and it would be absolutely true, but that's the whole point - the thing that gets you isn't digital vs. paper, it's doing the job well vs. doing it poorly.
>The goal isn’t to make theft impossible, just less profitable [...] all these technical measures are just noise.
Those two bits are in direct contradiction. The technical measures make the phone less valuable by preventing you from selling it as-is into the most expensive markets. That makes the theft less profitable.
You don't have to bring the profit to zero, you just have to push it as far down as you can. Every little bit will make the police work that much more effective.
Complex problems need to be addressed by multi-tiered solutions, where the problem needs to encounter the cheapest tiers first, and the most expensive tiers last.
Police is really expensive. We need police, and we need it to work properly, of course, but it can't be the first solution for every problem. Wherever there is a cheap and effective fix for 99% of a problem, you need to deploy that, and use police to deal with the remaining 1%. Otherwise, the economics just don't work.
Neglecting a simple technical fix while calling for a law enforcement solution, is like suggesting that we could all just dump trash in the streets, if the street cleaning crews just were more efficient. It just doesn't work. The street cleaning crews can only do their job if they are dealing with a tiny fraction of all trash. Police is the same. They can't deal with a society with strong incentives to being a criminal. Fix the incentives, then they can deal with the rest.
On the other hand, I really don't understand why the size of exfiltrated data seems to always be touted as a major index of the severity of an attack. One could steal several gigabytes that's a pre-release marketing video, or one could steal 1 kilobyte that's a .txt with the EW countermeasures master password in plaintext.
Wait. I'm not sure I'm getting the technical side of this, or if I'm getting it, I don't believe it.
A website that is not served by localhost is allowed to open arbitrary ports on localhost now? Surely that's impossible, as it would result in every PC in the world being breached in about ten minutes?
>Unlike ChatGPT which is an on demand thing that you have to prompt, on PCs it can be involved in what you are doing, and help you do it.
But you don't need an "AI PC" to do that. The app can just make remote API calls to an online LLM. There are several apps that work exactly like that. You'd only need an "AI PC" to do it offline, but that's a corner case these days.
The accountants still have to learn that "value is extremely hard to quantify" does NOT equal "value is zero".
Inspiring people has value. Hard value, even. Cash value. GDP value. It's just nearly impossible to put a number to it, but that number exists and is greater than zero, often greater than the easily-quantified number you spent to get it, sometimes a lot greater.
Yes, things that don't fit on a spreadsheet make the job harder. Well, tough. Time to earn your pay.
Make choices, of course, but you can't afford to ignore that profit outright, just because it doesn't fit on a spreadsheet. And I mean that "can't afford" in the most literal economic sense. Those who grab it will definitely put those who don't out of business. You see it happening all the time as uninspired companies grow creaky and eventually collapse. No reason to believe it can't happen to nations.
The problem is that member states are generally reluctant to devolve power to European institutions. That VAT exemption to protect your pet artisan cheese industry? The cheese industry whose union worked so hard to bring you votes? The VAT exemption that you so carefully crafted to make it evade state-aid legislation? EU-level fiscal policy would take it out overnight. There must be thousands of similar situations across the EU.
And campaigning against EU integration is really easy. Campaigning for the status quo generally is, because you don't have to explain what world you want, as it's the one we already have. So just draw up some memes and mockeries, never explain anything, and you're good. Campaigning for change is difficult.
Easy. Lack of jobs will drive economic decline and wage collapse, until hiring humans becomes cheaper than maintaining robots. The eventual equilibrium point of this process might be higher than the point at which low-skilled worker conditions and extreme inequality drive a series of bloody revolutions, but I wouldn't bet on it. Although you could, within limits, keep a lid on it through brutal repression.
Eventually, we'll just accept that we have to rework the economic system so that working becomes optional, although the rivers of blood might need to flow for a while before we get it.
>Communication matters, sure, but let’s not pretend we’re training novelists.
It doesn't sound like what we're trying to do. More and more, I'm encountering people who can technically read and write, but are unable to actually do this in a useful fashion. I waste sooo much time, on emails that consist of a screengrab and one line with no actual information on what the problem is, or on people who read an error message stating "the process cannot proceed because the engine on conveyor M23 is not starting" and then have to call me to ask what that means, or on meetings where somehow everyone agrees but it takes half an hour to figure it out. And it's been steadily getting worse. Doing something about this is probably a good idea.
>The "Developer" (I use the term loosely) couldn't figure how to do hours and minutes so used hours and decimal parts of hours (so 1 hour 15 minutes would be 1.25 hours)
Once I had to make my program produce a CSV that included some durations. The client wanted the duration to be human-readable, so that 1 hour 15 minutes should display as 01.15. The client then proceeded to open this in Excel, have it sum the column, and complain to me that the sum was wrong. -_-'
That's a fair point, but operation of the trains is tightly coupled to the rail system, and also limited by it. I don't think the network should ever be private, but I think private trains can work, with adequate regulation and strong enforcement. I don't think they would work efficiently with loose regulation.
>The problems we've seen in energy, water, and rail come down to a mix of ineffective regulation and calamitous government policies, rather than a failure of privatisation in principle.
Sort of. Natural monopolies handled by private operators can't have meaningful competition. That means they can, and do, fail in principle (e.g. by fundamental economic theory, rather than by individual bad actors). Unfortunately, there really isn't a self-adjusting system for operating natural monopolies, as free markets notoriously don't work for monopolies. I'm not saying that the state always handles monopolies well, but there's really no reason to expect the private sectors to naturally do better, except for ideology. They'll just suck in a different way.
>If we had responded like this at the start of the internet...
That's... no, just no. At the start of the internet, there was no web-wide tracking and mass scraping was only for indexing search. Despite that, we still had an explosive growth of services, many of them free to the end-user. There were message boards, there was free email, decent search services, everything. In some cases, it was better than what we had now. Even advertisers had it a lot cheaper, but still effective. Seriously, wth are you going on about?
>tech companies [...] only want to sell you things
No, they don't. Have you ever bought anything from Meta? Didn't think so. Almost nobody has. What the tech companies want is to sell ads to advertisers. I really can't believe that this is a point that still needs to be made.
In order to do that, what tech companies want from me is my undivided attention and full control of what I see, hear or think. That's the thing I want to fight, and since they are far too powerful for me to fight it effectively on my own, damn right I'm going to band up with other like-minded people and do something about it as a group. And in some cases, that "group" means "state". The rhetoric that says that states are inherently bad, unsurprisingly, originates from entities that very much want the little guys to fall divided.
From what I understand, the problem is that "legitimate interest" is too loosely defined. I think the original intent was that if an organization legitimately has your data (because that's how its service works), then there are some purposes for which it should be allowed to use said data, even if they are not strictly necessary to provide the service. Specific examples include fraud detection, IT security, and direct marketing (not reselling!)
I really don't think that the intent was that once an organization has your data, it should be able to use it for anything and everything it wants as long as it doesn't resell it. But that's how Meta, and everyone else, is trying to interpret it.
I think this can only be settled in court. I wish the law had been written with an explicit whitelist of allowed actions, but I guess that was deemed too inflexible in the face of future applications.
>But analysing that content, for any purpose whatsoever, is not the same thing. I cannot see how that could possibly infringe copyright.
But that's not how it works. Fair use doctrine is a blacklist, not a whitelist. If something is undefined because it's a new activity that doesn't resemble other allowed activities, then it's disallowed by default.
And no, LLM training does not look like reading at all.
>Home users don't want to be "bothered" by such things.
This. The fundamental issues with computer security are not going to be fixed, until the average Joe looks at a system with no password, and instinctively feels the same subtle wrongness you'd feel if you saw a house's main door with no lock.
>Or am I wrong?
Short answer: yes, you're wrong.
Long answer: the amount of energy humans consume is roughly five orders of magnitude less than the amount of energy the Earth receives from the Sun. If fusion worked, and then we shifted all energy production to fusion, and then everyone on Earth started consuming at USA-levels, and then everyone for some reason started consuming 10 times that - I don't know, flying cars, whatever - it would still increase Earth temperature by less than what random Sun fluctuations do. All of the scenarios where direct energy release becomes meaningful are so far-fetched as to count as high sci-fi (flying cities, things like that).
Climate change happens because greenhouse effects can trap vast amounts of Sun energy that would otherwise be reflected, and not because of direct energy releases.
>According to Microsoft, one of the most common frustrations is "finding and changing settings on your PC."
In fairness, that is one of the most common frustrations. Although my suggestions for addressing it would have been to (1) stop adding functions nobody wants, (2) stop changing the settings GUI with every version.
Oh, and (3) fire whoever came up with setting screens that are mostly negative space with a few settings here and there. I've just opened the screen settings; it's a grand total of 2 buttons, 5 combo boxes, 3 check boxes, and the monitor arrangement thingy, and yet somehow I have to scroll to see them all. At fullscreen 4k. While three quarters of the screen is empty space. That's insane.
Okay. When are we going to look into how social media harms adults? Or have we already established that turning 18 flips the immune-to-bullshit switch, so everything magically becomes harmless?
I don't want them banned, but I do want someone to look into this. And to do it properly, not do one study following 16 people for three weeks and then forget about it.
>Forgotten covid and the suppression of any alternative views?
What suppression? Social media were chock full of people claiming that COVID was just a flu and that vaccination was too risky, and those were just the coherent ones. In my country, some of those were on freakin' national TV every other day. Sure, they were roundly debunked and called "fake news" and misinformation, because they were, and they weren't invited to speak everywhere, because nobody is owed that. Frankly, I would have much preferred for at least major TV channels to just stop inviting them, but I respect their editorial authority. As far as I know, nobody was ever prevented from speaking in a public place, and nobody was prosecuted for social media posts.
Or do you mean that having someone contradict you in public is "suppression"? Or that not being allowed on private events that don't want you to attend is "suppression"?
Because those line of thoughts, now those really are suppression.
>I thought GOP/DJT were in favor of devolving to states?
They aren't. I don't know whether there ever was a time when the central-government-versus-local-government axis was represented in political splits, but today no party is ideologically in favor of either central or local. They just make different choices on which bits of policy ought to be centralized or not. That tends to depend on whether the current local managers for that policy are aligned or not.
Conservative parties claim to be in favor of local, but it hasn't been true for a while. Trump has been signing order that crap over states left and right, and as far as I can tell it's not much different in other countries. Here, one of the main right-wing parties has been claiming to be strongly in favor of devolving power to local governments for nearly 30 years, and yet the major actual reforms in that direction have been enacted by left-wing governments. Meanwhile, the current right-wing government is doing things like trying to dictate school and health policy, which have been devolved powers for a long time.
>Does this publication and the lefty contingent want censorship, loss of freedom and to become slaves?
People are getting grabbed off the street and detained for writing the wrong thing. Tourists are actually scared of visiting. I myself always wanted to visit the USA, and I'm not going to do it until the situation changes, because I have criticized Trump on social media and I'm scared of what immigration personnell might do to me. That's not the same thing as having a Facebook post removed. It's not even the same game.
It's not the concept of tariffs that's bad. That's a stupid claim that Trumpsters like to try to put in their critics' mouths; a strawman.
The problem is wide-area tariffs that have no specific purpose, and unpredictably shifting tariffs. The first just plain don't work except to depress everyone's economy; this has been seen in practice enough times and really shouldn't need another experiment to prove it. The second fail at shaping companies' behavior, because they can't plan to accomodate them; I'm not going to commit to a decade-long investment based on tariffs that might not be there next month. Trump has managed to do both at the same time.
If Trump had implemented targeted tariffs with a declared purpose and stuck to them, there would be far less to criticize. Most countries do some of that, here and there.
>you could easily build and pay for metro systems in basically every single major city in the USA.
No, you couldn't. It's not about money. There are not enough digger machines, there are not enough factories that build digger machines, there are not enough train manufacturers, there isn't enough trained personell to operate all of that, and nobody is going to build/train those on the assumption that Trump won't change his mind the next month.
You could build metro systems in every single major city in the USA by throwing money at the problem, but you can't do it in a President's mandate; you'll need a program that everyone can count on to remain stable for decades. "Stable for decades" is exactly the opposite of what's going on.
The world does not run on money, it runs on trust. And Trump is bankrupt.
Nobody said that you shouldn't boost local manufacturing, but this doesn't mean that running around naked and screaming should be lauded because it's "doing something".
First of all, you can use tariffs for this, but they have to make sense. You may notice that both the Biden administration and the EU levied tariffs against China, and while there was a lot of debate, as there should be, there was no massive disruption because of this. Make them targeted, make them gradual, couple them with local incentives, and so on and so forth.
Sure, this will take a long time to have an impact, but how could it be otherwise? Building capacity and restructuring supply chains takes years, building expertise for specialist sectors takes decades, and you still need widgets in the mean time. How could there be a quick solution?
Secondly, at this point there is pretty much no way for Trump, specifically, to make this work. The reason is point number (3) above. Building a factory takes years, turning a profit on it takes longer. Trump has changed the rules multiple times in a span of weeks, he is only there for three years and a half, he has made no attempt to make this even slightly bipartisan in any way, shape or fashion, a lot of people within his own party don't like this anyway, and he has already shown himself untrustworthy on his own trade deals.
If he said tomorrow that, look, sorry for the confusion, but here are the definitive tariffs; they will not change for the rest of my mandate, and they are unlikely to be changed quickly even if the democrats win in 2028, why on Earth would a business owner believe him, based on what we've seen so far? If I get a loan to build a resin mould production line and then Trump changes his mind and lowers China tariffs to 30%, I'll instantly go bankrupt. If Trump keeps still, but then the following administration does something similar, I'll go bankrupt in four years. I'd be insane to take that bet. Hell, the bank would be pretty reckless if they gave me that loan.
At this point, another President might find a way to do it, but Trump, on this, is already burned. He doesn't have enough time to regain sufficient trust.
>If *only* there were an American print outfit. If *only* companies could print on paper without resorting to China. Alas, clearly that's impossible, will never be possible, and to suggest such a thing is a direct, personal attack on those corporations who can't exist without China.
You haven't read the article. Or, if you've read it, you've chosen not to understand it, in order to further your point. The people interviewed have very clearly explained why their problems cannot be fixed by shifting manufacturing to the USA. I'll synthesize the main problems, for anyone else reading (I don't think you'll pay attention).
1) Sure, you have printers in the USA. Do you also have time machines? They are getting taxed on orders which were placed many months ago. Some of those bills are crushing.
2) It's not just printers. For many game parts, the relevant manufacturers either do not exist in the USA, or they are so vanishingly few that they have nowhere close to sufficient capacity. Setting up a new small batches mould production line takes many months at best, a year or two at worst. What exactly is your small business going to do in the mean time?
3) Those production lines, once set up, take many years to pay for themselves. How long are the tariffs going to stay? Long enough to turn a profit on your new USA production line? Would you bet your whole business on it? Yeah, thought not. Because of this, the capacity that's not there will not appear any time soon.
4) Even after everything is sorted, the price of the USA-made finished product would be much higher, because there's no way manufacturing in the USA is as cheap as in China, not even close (unless something goes even more horribly wrong). You okay with paying 100$ for a basic Monopoly box? 250$ for Hero Quest? 500$ for Descent? Yeah, I thought not.
The most likely final outcome, unless something changes, is that a crapload of producers will fold, and those who survive will make games without minis and counters, and just make everything from cardboard, while still pricing them the same or more. Sure, you can say screw it, it's just games, and back when you were a kid you had so much fun with cardboard tokens anyway. Now project this to some sector of the economy you actually care about...
It's more tragedy of the commons, than cognitive dissonance. Coordinated action would fix the problem for everyone, but individual action just puts the individual at a competitive disadvantage without fixing the problem for them at all. This means that, without the ability to coordinate, the problem cannot be fixed.
Unfortunately, there are some political forces that draw benefit from preventing coordination, and some powerful economic actors that benefit from not fixing the problem, and have figured this out sooner than everyone else. The relentless focus on individual action, which, while good and useful, by itself will not fix the problem, is a symptom of this.
It is no coincidence that British Petroleum came up with the "carbon footprint" concept. Anything to distract from the fact that appeals to goodwill cannot fix this; there are far too many disincentives.