iPhone keyboard is probably one of the several biggest factors I consider when once again I think "hmm, maybe this time I should upgrade to iPhone?". And then I'm confronted with this, ummm... thing, and immediately remember why I ditched iPhones years ago :) . How do you deal with it daily? I'm at a loss really. PS: I've owned 3GS and 4S and a few iPads, so I'm not just baseless here.
Exactly the same here. I've considered switching from Android multiple times and the two things that always stop me are notifications and the keyboard.
No long-press punctuation, no switch.
I also can't trust Apple to let 3rd party keyboards work smoothly everywhere, so that's not really an option I'm willing to take the risk on.
I switched from iOS to Android and it's actually kind of insane how much higher quality a lot of parts of the OS are. I wasn't expecting this, I was expecting customization and the associated jank. But no, it's been pleasantly surprising.
There's still some jank. Sometimes searching for something in setting takes upwards of 5 seconds. I can only assume it's downloading a bitcoin miner or something.
Someone can build a server in space, pairing a puny underpowered rack with a handful of servers to a ginormous football field sized solar panel plus a heat radiator plus a heavy as hell insulated battery to survive being a planet shade every hour for tens of minutes. We can do that from existing components and launch on existing rockets, no problem.
Why though?
Why would anyone need a server in space in the first place? What is a benefit for that location, necessitating a cost an order of magnitude higher (or more) compared to a warehouse anywhere on the planet?
I think we should split definition somehow, between what LLMs can do today (or next few years) with how big a thing this particular capability can be (a derivative of the capability). And then what some future AI could do and with how big a thing that future capability could be.
I regularly see people who distinguish between current and future capabilities, but then still lump societal impact (how big a thing could be) into one projection.
The key bubble question is - if that future AI is sufficiently far away (for example if there will be a gap, a new "AI winter" for a few decades), then does this current capability justify the capital expenditures, and if not then by how much?
Money can buy quality of life, and people earning 100-150k in USA per person in household do confirm this. And this purchasing ability is not linear, because of the fixed costs for many good and services. Previously many countries with low salaries had corresponding low cost of life (and cost of quality of life), but today the costs are rising faster than salaries everywhere across the globe, so the biggest winners are people who earn more in absolute values, hence rich Americans.
It actually can't, not generally at least for US labor.
One of the most important measures of quality is work-life balance. Basically, your life kinda sucks if you work all the time, and then you also get fat and sick and die young(er).
People in the US work a lot, and often the more wealthy, but not most wealthy, work A TON. In programming, it's not atypical to have "superstar" staff engineers putting in easily 60-70 hours a week. Of course, not including the commute.
But then there's the time off. Oh, where to begin. We're at a point where 10 days of PTO accrued a year is considered decent. It's work work work, and you can put in 20 years of service... and get, like, an extra couple days. Maybe.
None of this scales down. For example, I'm supposed to be working 40 hours a week. I'm not of course, the baseline is 45 because 9-5 is actually 9-6. And I haven't left at 6 in at least a year, so even that is underestimating it. But suppose I do work 40 hours a week.
Would I take a 50% pay cut to work 20 hours? Fuck. Yes. Yes. In a heart beat. But I can't, I'd actually be taking an 80% pay cut if I do that, so I couldn't live. And it's like this for literally ALL jobs. I can't just "move up", because the work-life balance doesn't get better, it actually just gets worse! And at no point can I take a "step down" and work less, because then I'm flipping burgers.
You honestly believe that a combined Germany-Poland-Baltic army, maybe with Italy's help and absolutely no USA involvement and manufacturing is today a viable threat to ruzzians? And that such an army is somehow more capable than today Ukrainian army in an all-out land battle with combined forces, permanently fighting for a decade now?
Of course such coalition has a big number of ultra expensive and effective weapons like planes, ships and tanks. That number of weapons will last for 3 months or so. Then what? Ruzzia is not a Taliban or Hamas, you can't just bomb them with impunity. Even half a century old soviet SAMs are valid threat to anything in the air, let alone newer ones. Plus Ruzzia is not alone, they have whole Axis manufacturing power potentially behind them - Iran, China, NK etc.
I would be very concerned about Ruzzia, if I were you. Just a thought experiment, what would Germany do when Ruzzian force will appear on the Poland-Lithuanian border, annexing all Baltic states?
I would like to point out two fundamental misconceptions towards the end of the article:
1. "AI is a filter. It strips away everything that can be automated, leaving only what requires actual thinking: creativity, collaboration, real-world problem-solving."
At any reasonable school-college level this is false. LLMs are perfectly fine replacing creative and collaborative aspects of work or study at intermediate levels. Yeah, they are unrealiable in results, but the participants don't care. LLMs produce data in shape of creative and collaborative work and that is enough to submit it.
What I mean by this, is when the house of cards will burn down (and it won't, because education is not a purpose of a school), it will also burn down creative training too.
2. Solving real/actual/applied/etc problem at the education facility of any level is nice and all (running business or solving community problems), but some studies just don't afford to be that "real" or "applied". All humanities, most of the harder STEM, etc. In USSR we had a whole separate class of higher-ed facilities, below universities called "technicums", which focused only on the applied knowledge. They were fine, but they definitely didn't fill all the demand.
That's fine, good even. Afaik at least for some of these tasks dev teams are doing a lot of manual tuning of the model (rumored that "r in strawberry" had been "fixed" this way, as a general case of course). The more there are random standalone hacks in the model, the more likely it will start failing unpredictably somewhere else.
That (freedom of payments) may have been the idea. But there are two problems with it:
1. Payments which you can't make today inside a legal system are two types. And if you enable system you automatically enable both types. For libertarians that is a clear 100% positive. For regular centrist citizens, not so much. At minimum it's a topic for debate.
2. BTC and a few other tokens actually make this problem worse. Since blockchain is public, you can always trace "bad" or real bad payment to the source wallet. That i one issue, and another is that since BTCs are non-fungible, the tokens used in such payments are forever tainted. Even in the current anarchic and almost unregulated environment some exchanges are blacklisting some of the tokens, to limit own exposure.
> Payments which you can't make today inside a legal system are two types. And if you enable system you automatically enable both types. For libertarians that is a clear 100% positive. For regular centrist citizens, not so much.
The problem with this argument is that cryptocurrency now exists whether it's legal or not and using it for illegal things is already illegal. Drug dealers are committing a felony by selling drugs and if that hasn't deterred them then neither will making something else they're doing a crime too.
So all of the negative uses are going to happen regardless of whether you also ban the positive uses. At which point, what are you gaining by making it illegal or inconvenient for innocent people to use it for something that isn't otherwise illegal?
> Since blockchain is public, you can always trace "bad" or real bad payment to the source wallet. That i one issue, and another is that since BTCs are non-fungible, the tokens used in such payments are forever tainted.
People keep making this claim and it keeps not making sense.
You don't need someone's permission to send them Bitcoin. Meanwhile large exchanges keep billions of dollars in a single wallet and have single wallets that do billions of dollars in transactions over a short period of time.
So let's consider the two possible ways this can work: First, if you get coins directly from a tainted wallet then you get in trouble, but if it was several steps back then it doesn't matter. This is, of course, useless, because then people would just transfer the coins through a couple of other wallets first.
Second, any wallets that receive any tainted coins become tainted forever. Then immediately the vast majority of the chain is tainted because the coins have made the rounds through a large exchange at some point. Worse, it's pointless to try to defend against it by refusing tainted funds, because you can't actually refuse funds. Your billion dollar wallet or freshly mined Bitcoin can be tainted by any troll who sends you a dollar from a tainted wallet without your permission, and trying to treat coins as non-fungible is probably a good way to get someone to troll you like that.
Which gives you two alternatives again. The first is that all coins can be tainted by trolls, which will in practice cause exactly that to happen and thereby make the premise meaningless. The second is that you can try to say that it doesn't count if someone sent them without your permission, but now you can't tell if something is tainted by looking at the chain because it can't tell you which transactions were unauthorized by the recipient, and moreover you would then have a mechanism for getting dirty coins into a clean wallet.
In other words, when anyone can send you money without your permission, your options are "everything is dirty" or "everything is clean".
> At which point, what are you gaining by making it illegal or inconvenient for innocent people to use it for something that isn't otherwise illegal?
The problem is scale. The more widespread is such system, the lower is the barrier to entry and the higher is cost to actually prosecute users to their amount and rate of usage (which we already see today).
Also this whole legal/illegal divide is often presented as if there was approximately same order of magnitude of both users. While I guess that actually the illegal use is way way larger than the legal use, simply because it is so crude and slow and buggy and unsafe by design. (excluding gambling, since that use is kinda derivative, depending on the all other uses making up a base on which to gamble)
And this is why token systems by rights should be heavily restricted, since they are so disproportionate in impact. We can all legally buy a knife in any shop, despite the fact that if used for attack a knife almost inevitably produces at least one body. Small arms are also available almost anywhere but with a lot of restrictions. Big arms are almost never available for purchase, just like explosives. And then the stuff like a canister of zarin is totally out of the question. That's because of the disproportionate effect. Same with financial instruments. Tokens are an Abrams of the finance world, and currently we let anyone have one, which is mindboggling to me.
> In other words, when anyone can send you money without your permission, your options are "everything is dirty" or "everything is clean".
You are correct. Afaik all tries to ban Tornado laundered tokens were eventually dropped. But the mechanism and potential still remains.
Also, please correct me if I'm wrong, in the case of BTC specifically we can track tokens from the "dust" attack and separate them from the legal and nice tokens, since they will stay in the different UTXO in the same wallet. Though I'm not very familiar with that, if it possible to pick which UTXO to transfer selectively.
reply