You say it requires a little acclimatisation, but the caller may well just not know anyone with that specific accent/have never had the chance to acclimatise to it other than this phone call, there's an awful lot of accents in the world.
Additionally I find even though I can understand an Indian accent for example quite well in person, I really struggle with it over the phone due to the compression causing quite poor sound quality and lack of facial expressions to be able to read (which I would be using in person to help me understand a strong unfamiliar accent), whereas when accent is more familiar to me, the poor audio quality and lack of body language isn't nearly as much of an issue, presumably as I just have way way more exposure to the accent so can fill in gaps better.
You don't need to accustom yourself to every individual accent; you need to practice hearing people say things in different ways. Once you are comfortable with some voices that are different to yours, it's much easier to understand other differences as well.
I'm sympathetic to audio quality issues. No one would object if they developed tech to improve call quality, but they didn't.
I think you’re not taking into account that there are people from hundreds of countries out there whose native language is not English, that are not exposed to different English accents at all other than watching movies etc, and suddenly having to deal with weird accents.
The whole thing with GPL code seems like a mess and surely couldn't be set as actual precedent, right? It is totally infeasible for me to check every single GPL project on every code hosting platform to see if the code Claude etc produced is too similar. If a set of training data used for the model was released to check against that would be one thing, but you can't honestly expect someone to check every repo available from all time to see if a model (that you are not informed of what it was trained on and therefore could reproduce) might've reproduced code from it.
That's not at all like checking the dependency chain of a dependency or anything as you can just read the licence of anything you're choosing to use. Surely the precedent would have to be that a model trained on GPL code has itself been infected by GPL, and therefore must have all source/weights released too if the assumption here is that it can have embedded the code well enough to be able to reproduce it?
> Surely the precedent would have to be that a model trained on GPL code has itself been infected by GPL, and therefore must have all source/weights released
I don't see how this follows, unless we also agree that humans who have ever read any GPL code are themselves permanently tainted and therefore cannot produce anything that isn't influenced even slightly by said code.
Is it just because we think the robot does a better job at learning than we do? It's an impossible line to draw, I agree, but I don't agree that the answer is "well then everything must be considered tainted," I say the answer is "ignore a vestigial concern of a bygone era."
The robot does a better job at reproduction. I don't think there exists a definition of "learning" unambiguous enough to make the claim that it learns better than humans. Specifically, published models don't learn at all -- after the training phase, the model weights are fully static.
Duplicating BSD-licensed code without copyright attribution and mention of the original license is just as much a violation of the original copyright -- that applies regardless of additional copyleft requirements imposed by the GPL. A different but no less serious restriction applies to all the code examples on MSDN: the license disallows using the samples in production code.
LLMs are effectively copyright laundering machines, and barring any indemnification clauses in the ToS (of course there are none), full liability lies with the user.
> It is totally infeasible for me to check every single GPL project on every code hosting platform to see if the code Claude etc produced is too similar.
I would say that choosing a tool that makes it infeasible doesn't actually excuse you from doing it.
> but you can't honestly expect someone to check every repo available from all time to see if a model [...] might've reproduced code from it.
Well, if you care about not violating any licenses, you could buy services from an LLM provider that was only trained on code in the Public Domain (or code that the LLM provider licensed for that purpose), and/or buy some kind of legal guarantee from the LLM provider that the code produced is "clean".
Of course, that'd be much more expensive than current offerings, but it would reflect the real cost of software development, not just YOLOing it, from a legal perspective.
When I wrote a book, part of the contract with my publisher was that I had to attest that I actually wrote the book myself, that quotes were properly attributed etc. If you buy code-writing services, why shouldn't it contain similar clauses?
It sounds like the real solution to this is to be able to control permissions at an OS level for network per app, as you would be able to do if you had root access. I have no idea why regular Android distros don't allow you to do this, it seems like a really sensible thing to expose in app settings given the permissions model of Android.
The difference I see there is any line in a C codebase could have these issues, whereas in Rust they're specifically marked as unsafe sections, with the language having a clear list of invariants it expects the programmer to uphold in an unsafe block. Additionally Rust has a culture of developers specifically justifying the unsafe block and why it's correct in comments. It's a massive reduction of the scope of the code that needs audited, that doesn't mean there can't be mistakes in it, just that it's easier to verify certain properties of the whole codebase.
What all these breaches tell me is that personal data should not be required, and especially not stored unless absolutely necessary. I cannot verify how my data is treated once it leaves my device, so how can I possibly trust it will be treated properly and not leaked?
This is a major reason as to why I am so strongly against all this verification shit governments keep trying to push, the best way to keep data secure is not to have it in the first place, therefore my personal data should not leave my device except in the strictest of circumstances for things like my name/DOB/address/SSN.
I find that a lot less problematic as there's just very few shorts on my feed, I've never been able to scroll through more than 5 or so without just going into ones I've seen before.
Additionally I find even though I can understand an Indian accent for example quite well in person, I really struggle with it over the phone due to the compression causing quite poor sound quality and lack of facial expressions to be able to read (which I would be using in person to help me understand a strong unfamiliar accent), whereas when accent is more familiar to me, the poor audio quality and lack of body language isn't nearly as much of an issue, presumably as I just have way way more exposure to the accent so can fill in gaps better.
reply