>what if the LLM gets something wrong that the operator (a junior dev perhaps) doesn't even know it's wrong?
the same thing that always happens if a dev gets something wrong without even knowing it's wrong - either code review/QA catches it, or the user does, and a ticket is created
>if it fails here, it will fail with other things, in not such obvious ways.
is infallibility a realistic expectation of a software tool or its operator?
Ballpark math says you could sustain it for half an hour on Hetzner for $5k-$6k (only from 1500 IPs though), at least if your account didn't get banned first and you're halfway decent at network programming. I have no idea what a proper botnet like this costs though or how large the profit margins are.
Yeah I assume there's the initial startup cost of successfully managing to infect a large network of devices, and then the cost for any given use is likely "what customers will pay for it". If they are selecting out big money targets and focusing on gaming, I'm guessing the price isn't that high, but they also presumably know interesting a state actor in taking them down either by changing targets or bringing in enough money is bad for business.
The idea is, the botnets are in control of someone else. Who "owns" them. And some of those will rent "their property" for money, like they would legitimately own them.
Depends. The more the owners use their bots, or let others use their botnets, the more attention there is to them and the less useful the botnet is (either blacklisted IPs or owners noticing).
And a little bit of malicious bandwidth is easy to hide, a lot not. So there is a price to bandwith to the criminal owner.
back in '98 i got a 100mb per download limit for $100 on my cable connection. i recall getting DoS'd by someone cause i was a lpb barstard in quake tf. They were kind though, only DoS'd me 90mb as a warning.... Years later, TF2 is getting DoS'd into oblivion, an extorhted by DDoS for hire. Some things change, some things stay the same.
I wonder how well that conversation detection works. Does it really help in a loud environment?
As a neurodivergent person I lack the innate human skill to filter voices out of a cacaphony of noise so loud bars etc are hell. There also the "talking with earphones in is rude" but that's an issue that can just be explained.
Needing root to enable it is a major deal-breaker though :( and moving to an iPhone is impossible for me. Too much stuff that's not supported.
Or Apple just doesn't want to bother with the nightmare of supplying and supporting an app to do all those things on other platforms, and in particular, there are regulatory approvals around the "hearing aid" feature that would pretty much require a specific device.
They have a basic app for some of their other devices like the Beats line. One other thing you simply can't do without pairing AirPods with an Apple device is enrol them in AppleCare One.
You're commenting on a post where a random guy provides this "nightmare of supplying and supporting an app" in his spare time, except he actually has to work around Apple's malicious obfuscation and standards non-compliance, so it would actually be way easier for Apple to do it themselves.
From what it says, a rooted device is required because Apple made them behave differently depending on the host. Apple wouldn't have needed a rooted Android device to support all the features.
Are you saying this would the first time an unpaid open source effort has done something a big company declined to do because of the operational costs they face?
It is in fact significantly harder for Apple. Because nobody expects random spare time GitHub project to work perfectly. Or even very well. Apple’s reputation, and trillion dollar market value, is based on the idea that their stuff works perfectly.
good god man, just accept that this is objectively an EXTREMELY easy thing to do for anyone. Yes theoretically there are things that are easier for OSS devs than large companies, THIS AIN'T ONE OF THEM.
Ugh, trillion dollar market value doesn't mean they are incapable of making a basic android app. Check their move to ios app if you have any doubts.
It doesn’t matter how frustrated you get or how many times you write capital letters, Apple is a private company and can do exactly what they want to do. If you would like Apple to do your bidding, acquire a controlling interest - it’s public so there’s nothing stopping you.
Welcome to this argument which is about how easy/hard it might be for a company to implement this particular feature.
The argument about whether they ought to is in some other thread I imagine, you might have lost your way. I don't own their airpods so in this particular instance, IDC about the outcome.
Are you aware they are maintaining multiple complete OSs, and multiple versions of each? With drivers for hundreds of components? The scope of AirPods on Android is laughable in comparison.
You're responding in a sub-thread where others have specifically called out the fact that you can't get battery status from AirPods on non-Apple platforms. This is, to my knowledge, a feature that is supported natively by the Bluetooth stacks on every mainstream OS and requires no "apps" at all. For example, I can connect my Bluetooth mouse to my Linux machine and it happily reports the state of the battery.
Care to offer a justification for why this is the case without resorting to "the multi-trillion-dollar behemoth can't be bothered to build an app"?
The multi battery levels thing is native proprietary on every platform since there is no Bluetooth spec for more than one battery level and even that just uses uint8.
As I posted elsewhere in the thread, this is incorrect. The Bluetooth Battery Service spec allows for a single device with multiple batteries and individual battery reporting for each. [0] They even give the example in that doc of earbuds which are one “logical device” but two physically separate pieces, each with its own battery.
As additional evidence, there are "AirPods-like" earbuds on the market such as the Sony WF-C700N, which have no problem reporting three battery levels over standard Bluetooth on e.g. Linux.
The Bluetooth Battery Service spec allows for a single device with multiple batteries
As of version 1.1 of the battery service which was finalized at the end of 2022. Given Bluetooth's track record, who knows what kind of interoperability landmines exist.
If you don't like the Apple device, use something else. It's not like a messaging platform where you'd need compatibility with other peoples' phones.
If you'd bothered to dig into the spec, v1.0 basically says do what you want. v1.1 defines a proper namespace and well known descriptions for multiple batteries. Apple did well to avoid the interoperability minefield.
> If you don't like the Apple device, use something else. It's not like a messaging platform where you'd need compatibility with other peoples' phones.
I own and use lots of devices, for both work and personal tasks, including Apple and non-Apple devices. I own a pair of AirPods. I'd like them to work well across all the platforms that I use. There is nothing technically preventing Apple from achieving this, aside from Apple's arguably illegal tying behavior.
> If you'd bothered to dig into the spec, v1.0 basically says do what you want. v1.1 defines a proper namespace and well known descriptions for multiple batteries. Apple did well to avoid the interoperability minefield.
I have read the spec; please don't accuse me of not reading it. Have you written Bluetooth device firmware before? In case you haven't, at a high level:
* The BT device exposes a "profile," which defines one or more "services", which are essentially different types of data that can be read from or written to the device.
* Multiple instances of the same type of service (the Battery Service in this case) can be exposed in the profile. I don't know if this ability was always present in the spec or was added after the fact, but it was, at minimum, present in 2011 when the BAS 1.0 spec was released.
* So, if your device has more than one battery, its profile will have an instance of the Battery Service defined for each one.
I will grant that the 1.1 spec document is a lot clearer and provides lots of diagrammed examples, but the only net new functionality in 1.1 are a set of new battery-related fields (these are called out near the beginning).
When a device has more than one instance of the Battery service, each Battery
Level characteristic shall include a Characteristic Presentation Format
descriptor that has a namespace/description value that is unique for that
instance of the Battery service.
1.1 says:
When a device has more than one instance of the Battery Service, each Battery
Level characteristic shall include a Characteristic Presentation Format descriptor
(Volume 3, Part G, Section 3.3.3.5 in [1]) that has the Name Space field set to
”Bluetooth SIG” and the Description field set to a valid value from the GATT
Namespace Descriptors [4] and that is unique among all instances of the Battery
Service exposed by the GATT Server.
1.0 was a mess and your anger over a poorly defined and relatively minor feature seems quite misplaced. Bluetooth interoperability has historically been a mess (still is from my experience). But go ahead be big mad that Airpods only play audio from third party devices and don't provide battery status in a way that adheres to a recent revision of the standard. Meanwhile I'm sure Sony would never use a proprietary format ever…
I had posted a reply addressing your points, but I don't think this discussion is productive and you don't seem to want to engage honestly with what I'm saying and stay on topic. So I'll just say have a good day.
So they refuse to report anything useful rather than make use of the single battery level. Amazingly every other brand of Bluetooth earbuds manage to report a useful battery level despite them having a separate battery in each side.
The Bluetooth spec only supports one battery status. AirPods have three batteries. Is 1 < 3 a satisfactory enough answer to you?
On the subject of the multi-trillion-dollar behemoth, Apple is a private company. If you have the capital, you can acquire a controlling interest and then they’ll work on whatever you like. Until then, you’re out of luck.
> The Bluetooth spec only supports one battery status. AirPods have three batteries. Is 1 < 3 a satisfactory enough answer to you?
No, it's not. The Bluetooth Battery Service spec allows for a single device with multiple batteries and individual battery reporting for each. [0] They even give the example in that doc of earbuds which are one “logical device” but two physically separate pieces, each with its own battery.
> On the subject of the multi-trillion-dollar behemoth, Apple is a private company.
Apple is, by definition, a public company.
> If you have the capital, you can acquire a controlling interest and then they’ll work on whatever you like. Until then, you’re out of luck.
No. Anticompetitive behavior such as tying (what I would argue is happening here) can and should always be subject to examination, criticism, and possible litigation by the public.
Always this sad argument that X is a private company and they can do what they like.
Companies are not acts of God or nature. They are a private company operating on a society that allows it to exist because it is believed to be the for the public good. The public has very much the right to question it's practices, and if they are anti consumer, monopolistic, or a list of other things, to correct them. Shareholders be damned.
So what's your argument then? Companies can't release a product unless each and every feature works with their competitors products? By that logic most of the software and hardware you use today simply would not exist.
Like a lot of parts of the (especially earlier revisions of) Bluetooth spec the battery status took a slapdash approach to defining things. Look at anyone who's used Bluetooth on Windows to see what a nightmare interoperability still is. So Apple released ear buds that implement poorly defined parts of the spec but otherwise work with third party bluetooth devices, and that's bad?
Yikes.
Meanwhile, the Bluetooth SIG released an update at the end of 2022 that actually starts to require some sort of standardization. You know who's name was on that little update? Big bad awful anticompetitive Apple.
Yeah, there are two batteries, the one in the earbuds and the one in the container. There's no way in BLE to transmit both values - and choosing either one is lying to the user about something.
It's not uncommon (at least for me) to have a low earbud battery level (because I've just binged Slow Horses) or a low container battery (because I've just charged the earbuds from the container for the third time and drained the container). There's a suggestion above that you should "just choose the lowest one because 99% of the time that's what you're interested in", except that's not true in the second case.
I'm fairly sure that if you could report both, then Apple would report both using this hypothetical standard method, but since you can't, and there's no easy way to just "choose one" without misleading the user about something, they choose to do it properly, even though that means it's an Apple-only thing.
See my other replies in this thread — it’s totally possible to do with standard Bluetooth, yet Apple doesn’t do it. So your “fairly sure” assumption that Apple would make use of this feature if it existed seems to be wrong.
What "other platforms" are you talking about? Just an Android app would suffice. It's not a huge deal for a company worth trillions, especially if the features are already there and they're just blocking non-Apple products. If they deliberately do that, it makes you think they don't really care about their customers and are more interested in locking people into their ecosystem.
>Some skills, like framing, values, balance, etc. become even more important differentiators.
I agree. I think many artists in the future will be closer to directors/cinematographers/editors than performers
Many of the skills artists have today will still be necessary and transferable, but what will separate the good artists from the bad artists will be their ability to communicate their ideas to agents / other humans
Same with software developers I suspect - communication will be the most important skill of all, and the rockstar loner devs who don't work well in teams will slowly phase out
>AI has no intent or creativity, so it can be neither right nor wrong, neither good nor bad.
AI is just a wrapper around a tool - it doesn't need intention or creativity because those come from the user in the form of prompts (which are by definition intentional)
It's just a Natural Language Interface for calling CLI tools mostly, just like how GUIs are just graphical interfaces for calling CLI tools, but no one thinks a GUI has no intentionality or creativity even when using stochastic/probabilistic tools
Anything a user can do with an AI they could also do with a GUI, it would just take longer and more practice
>Either everything generative AI creates is slop or nothing is. So everything is.
But then how do you know something is slop before you know if it's made with GenAI? Does all art exist as Schrodinger's Slop until you can prove GenAI was used? (if that's even possible)
the same thing that always happens if a dev gets something wrong without even knowing it's wrong - either code review/QA catches it, or the user does, and a ticket is created
>if it fails here, it will fail with other things, in not such obvious ways.
is infallibility a realistic expectation of a software tool or its operator?