But also they are just tuning and packaging a publicly available model, not creating their own.
So they can be profitable because the cost of creating that model isn’t factored in, and if people stop throwing money at LLMs and stop releasing models for free, there goes their business model. So this is not really sustainable either.
It’s popular because it appears in the opening of the anime Slam Dunk, which plays in the Shonan area.
No I haven’t watched the anime, I found out about it a couple years back when I was confused why there were so many people there.
The design was selected after the city government invited public submissions to promote Kamakura.
…
“We decided on the suspension because some residents thought the design helped attract visitors and found this unpleasant,” a member of the city’s tax division said.
I want to post the surprised pikachu meme but I actually have severe doubts that this is what’s attracting the tourists taking pictures at the crossing…
We need to start posting this everywhere else too.
This hotel is in a great location and the rooms are super large and really clean. And the best part is, if you sudo rm -rf / you can get a free drink at the bar. Five stars.
How do they mess this up so bad?
They made their devs use copilot.
In principle, yes, but isolated unvaccinated communities can still have major incidents just from one contact. E.g. Germany is at around 97% and a couple weeks ago there was a small outbreak involving a bunch of unvaccinated children.
It does include the fifth password Louvre4, but actually they only can’t re-use the previous 6 passwords so it went through.
I don’t understand how people can look at the insane progress gpt has made in the last 3 years and just arbitrarily decide that this is its maximum capability.
So this is not entirely arbitrary, and probably part of it is also that they’re not just looking at the progress, but also at systemic issues.
For example we know that larger models with more training material are more powerful. That’s probably the biggest contributing factor to the insane pace at which they’ve developed. But we’re also at a point where AI companies are saying they are running out of data. The models we have now are already trained on basically the entire open internet and a lot of non-public data too. Therefore we can’t expect their capabilities to scale with more data unless we find ways to get humans to generate more data. At the same time the quality of data on the open internet decreases because more of it is generated by AI.
On the other hand, making them larger also has physical requirements, most of all power. We are already at a point where AI companies are buying nuclear power plants for their data centers. So scaling in this way is close to the limit too. Building new nuclear power plants takes ages.
Another different thing is that LLMs can’t learn. They don’t have to be able to learn to be useful, obviously we can use the current ones just fine at least for some tasks. But nonetheless this is something that limits the progress that’s possible for them.
And then there is the entire AI bubble thing. The economical side of things, where we have an entire circular economy based on the idea that companies like OpenAI can spend billions on data centers. But they are losing money. Pretty much none of the AI companies are profitable other than the ones that only provide the infrastructure. Right now investors are scared enough to miss out on AGI to continue investing but if they stopped, it would be over.
And all this is super fragile. The current big players are all using the same approach. If one company makes that next step and finds a better approach than transformer LLMs, the others are toast. Or if some Chinese company makes a breakthrough with energy usage again. Or if there is a hardware breakthrough and the incentive to pay for hosted LLMs goes away. Basically even progress can pop the bubble because if we can all run AI that does a good enough job at home then the AI companies will never hit their revenue targets. And then the investment stops and companies that bleed billions every quarter without investors backing them can die very quickly.
Personally I don’t think they will stop becoming better right now. Even if they do stop, I’m not convinced we understand them well enough to be unable to improve the ways in which we use them a bit more. But when people say that this is the peak, they’re looking at the bigger picture. They say that LLMs can’t get closer to human intelligence because fundamentally, we don’t have a way to make them learn, they say that the development model is not sustainable, and other reasons like that.
I started learning when I was 9. I think to some extent it was easier back then in the 80ies because computers were relatively simple machines. On the other hand I also had to learn English at the same time to be able to read manuals and programming books etc. So I think it must be possible because even if I saw the word “syntax” I doubt I had a full grasp of what it means.
At the end of the day this is just another right wing conservative politician in the same right wing conservative party that’s been ruling Japan for almost the entire time since 1955.
It’s definitely a mixed vibes game :)
I think it’s decent as a roguelite shooter. But it’s a bit slow and you have to spend some time playing to appreciate it. For example I saw some reviews complaining that there are only four ranged and three melee weapons. Which is true but completely disregards how different the variants of each can be. For example there are normal grenade launchers but there also are launchers that shoot large plasma balls that remain where they land for a short time and do damage over time, there are launchers that shoot sticky timed explosives, etc., and then among those, some will shoot one at a time, some will shoot a whole set of bombs in one burst, some will shoot smaller projectiles but really really fast…
The melee weapons are a bit unbalanced though. Simply because the axe, the stereotypical slow heavy hitter weapon, can generally kill normal enemies in one hit and the other melee weapons, while not weak on paper, can’t do that. So quite often the axe is just the most convenient.
They did a good job making us use both ranged and melee weapons though with the energy system, and in addition we also have a drone with one out of several support functions, a shield, and dash, and dodge, and a super move for each weapon. So overall there is good variety in combat.
They did make some puzzling decisions too where it sometimes feels like they don’t understand the genre they picked. For a roguelite, the individual runs can be too long. If we beat a boss, we get a shortcut to that level, but as a consumable that can be used only once. At the same time the upgrades between runs take a lot of materials so skipping a few floors might not even pay off in the end. And if we die we lose some of the materials we collected, too (this can be eliminated with some upgrades relatively early, but who does this in a roguelite…) And the biggest one is just the price. For a roguelite it’s fairly expensive.
The setting is pretty far out too. For me a huge plus just because it’s not the same as every other game I’ve been playing recently. I play so many fantasy games that playing a cyborg who live streams shooting up a company in a dystopian future feels really fresh.
Full Metal Schoolgirl for Switch 2. Work’s been frustrating so the premise sounded intriguing
On Mac:
If you want an icon you can double click on your desktop, you can put you command in a file with the extension “.command” and mark it as executable. Double clicking it will run the content as a shell script in Terminal.
If you want something that can be put into the Dock, use the Script Editor application that comes with macOS to create a new AppleScript script. Type do shell script "<firefox command here>" then find Export in the menu. Instead of Script, choose export to Application and check Run Only. This will give you an application you can put in the Dock.
If you want to use Shortcuts, you can use the Run Shell Script action in Shortcuts too.
Finally, if you want something that opens multiple firefoxes at once, chain multiple firefox invocations together on one line separated by an ampersand. There is an option you have to use (–new-instance I think?) to make Firefox actually start a complete new instance.
That’s funny because I grew up with math teachers constantly telling us that we shouldn’t trust them.
Normal calculators that don’t have arbitrary precision have all the same problems you get when you use floating point types in a programming language. E.g. 0.1+0.2==0.3 evaluates to false in many languages. Or how adding very small numbers to very large numbers might result in the larger number as is.
If you’ve only used CAS calculators or similar you might not have seen these too since those often do arbitrary precision arithmetics, but the vast majority of calculators is not like that. They might have more precision than a 32 bit float though.
I mean, most calculators are wrong quite often
What bothers me the most is the amount of tech debt it adds by using outdated approaches.
For example, recently I used AI to create some python scripts that use polars and altair to parse some data and draw charts. It kept insisting to bring in pandas so it could convert the polars dataframes to pandas dataframes just for passing them to altair. When I told if that altair can use polars dataframes directly, that helped, but two or three prompts later it would try to solve problems by adding the conversion again.
This makes sense too, because the training material, on average, is probably older than the change that enabled altair to use polars dataframes directly. And a lot of code out there just only uses pandas in the first place.
The result is that in all these cases, someone who doesn’t know this would probably be impressed that the scripts worked, and just not notice the extra tech debt from that unnecessary dependency on pandas.
It sounds like it’s not a big deal, but these things add up and eventually, our AI enhanced code bases will be full of additional dependencies, deprecated APIs, unnecessarily verbose or complicated code, etc.
I feel like this is one aspect that gets overlooked a bit when we talk about productivity gains. We don’t necessarily immediately realize how much of that extra LoC/time goes into outdated code and old fashioned verbosity. But it will eventually come back to bite us.
I have to do a bunch of relatively unsurmountable steps to do what should’ve taken half a minute. Like screenshot the profile and scrape the text with iOS Photos text recognition.
The iOS workaround isn’t quite as unsurmountable as you don’t have to go through the Photos app at all. You can enter text selection mode directly from the screenshot without even saving it or leaving the app you’re in. And since iOS will look up any word you can select in the system dictionary and also translate any text you can select, you can do these things right there too.
That said I did once make a shortcut that lets me triple tap the back of my phone to pop up a text version of everything on screen that the iOS OCR detects. Not sure what I did that for though, I don’t really use it.
Well it’s not improving my productivity, and it does mostly slow me down, but it’s kind of entertaining to watch sometimes. Just can’t waste time on trying to make it do anything complicated because that never goes well.
Tbh I’m mostly trying to use the AI tools my employer allows because it’s not actually necessary for me to believe that they’re helping. It’s good enough if the management thinks I’m more productive. They don’t understand what I’m doing anyway but if this gives them a warm fuzzy feeling because they think they’re getting more out of my salary, why not play along a little.
What gets me is that even the traditional business models for LLMs are not great. Like translation, grammar checking, etc. Those existed before the boom really started. DeepL has been around for almost a decade and their services are working reasonably well and they’re still not profitable.
A startup spun out by a university (z.ai). Their business model is similar to what everybody else does, they host their models and sell access while trying to undercut each other. And like others they raised billions in funding from investors to be able to do this.