That's just it, it's not.
It's a whole economy based on the threat of neverending warfare.
An actual war, where they aren't just stomping on something from a great height, that is short term benefit compared to selling them the re-up every cycle, "just in case"
Military keeping up with the Joneses.
The China vs US one ?
Brigades are out in full force in those trenches
Greater compared to human code? Not sure about that, but I'm not disagreeing either. Greater compared to verified able programmers, sure, but in general?..
Both.
The reasons are quite hard to describe, which is why it's such a trap, but if you spend some time reviewing LLM code you'll see what I mean.
One reason is that it isn't coding for logical correctness it's coding for linguistic passability.
Internally there are mechanisms for mitigating this somewhat, but its not an actual fix so problems slip through.
I don't think I'm getting your point here. Do you mean by that, the code basically lacks focus on an end goal? Or are you talking about the fuzzyness and randomization of the output?
The latter, if you give it the exact same input in the exact same conditions, it's not guaranteed to give you the same output.
The fact that its sometimes close to the same actually makes it worse because then you can't tell at a glance what has changed.
It also isn't a simple as using a diff tool, at least for anything non-trivial, because it's variations can be in logical progression as well as language.
Meaning you need to track these differences across the whole contextual area which, if you are doing end to end generation, is the whole codebase.
As I said, there are mitigations, but they aren't fixes.
Let's assume we're skipping the ethical and moral concerns about LLM usage and just discuss the technical.
it makes an impression on me as if human code would be free of such errors
Nobody who knows anything about coding is claiming human code is error free, that's why code reviews, testing and all the other aspects of the software development lifecycle exist.
To me it sounds like nobody should ever trust AI code
Nobody should trust any code unless it can be verified that it does what is required consistently and predictably.
because there can or will be mistakes you can’t see, which is reasonably careful at best and paranoid at worst
This is a known thing, paranoia doesn't really apply here, only subjectively appropriate levels of caution.
Also it's not that they can't be seen, it's just that the effort required to spot them is greater and the likelihood to miss something is higher.
Whether or not these problems can be overcome (or mitigated) remains to be seen, but at the moment it still requires additional effort around the LLM parts, which is why hiding them is counterproductive.
At some point there is no difference anymore between “it looks fine” and “it is fine”.
This is important because it's true, but it's only true if you can verify it.
This whole issue should theoretically be negated by comprehensive acceptance criteria and testing but if that were the case we'd never have any bugs in human code either.
Personally i think the "uncanny valley code" issue is an inherent part of the way LLM's work and there is no "solution" to it, the only option is to mitigate as best we can.
I also really really dislike the non-declarative nature of generated code, which fundamentally rules it out as a reliable end to end system tool unless we can get those fully comprehensive tests up to scratch, for me at least.
Iirc the situation is similar in the UK, for hunting and "pest control"
By the sound of it, the disagreement is mostly in how direct an impact AB1043 will have on government plans for data collection and authoritarianism.
That's not really the original disagreement i was referencing, nor is it a position i've taken, we agree that the local only bill isn't the big bad.
You twice referenced the slippery slope fallacy when replying to comments clearly describing future actions, i was pointing out that it doesn't meet that criteria because there is a reasonable assumption that the described escalation will occur.
Your original responses to which i was referring:
This is a slippery slope falicy. Just because the option is provided to self-identify age, doesn’t mean that it will be replaced with more complex and direct data collection (which I am against, if it wasn’t clear) later
You’re again relying on slipery slope falacy to say that because I’m okay with this one specific form of age gating, I’m okay with every other one, which I have repeatedly made clear is not true.
The first one is the main issue i was pointing out, the second one isn't how the fallacy is applied at all.
As no one is taking the position that AB1043 is the actual danger most of what you are arguing doesn't really apply.
Similarly with the Overton window, where it has been standard practice for over a decade to have a “are you at least 18?” popup, and for every single service to ask you your age, if not more. We absolutely need more data protections for systems such as this (ideally an outright ban on saving this information) but this doesn’t seem to make it worse.
Emphasis mine.
Hard disagree, moving the responsibility of this from individual websites to the OS is a big jump in scope.
The same kind of jump as making it the ISP's responsibility if they serve illegal content from individual websites ( as has been suggested ).
Aside from that it centralises the surface area for future changes and enforcement.
Basically, from my understanding, this isn’t a step towards data collection or authoritarianism, and provides no significant benifit to either of those causes - its effectively a technical standard.
This is the disagreement, i (and obviously many others) are pointing at the long and comprehensive list of similar initiatives, both recent and historic, that were stepping stones to further encroachment and saying "oh look another small step in the continued and provable encroachment upon privacy" and you seem to be advocating for the benefit of the doubt.
Like, if this age-verification flag was proposed by the Linux Foundation, and agreed to by others, would the backlash be this big?
If the linux foundation had the same history of shenanigans, then yes.
Similarly, I don’t see any contradition between wanting a ban on storage/sharing of user data, and the implementation of a flag like this - even if we are able to ban all storage of user data, this law would be unaffected. That’s what I’m trying to figure out - how do people think that this leads towards those end goals? How would blocking it improve anything?
Ignore the technical implementation of this one step, nobody is saying this is the endgame big bad.
Think of it as a prevention measure, a single ant in the kitchen isn't a problem in and of itself, but it's almost certainly an indication of a larger potential future problem.
You are arguing it's not a problem because the ant only has 5 legs, everyone else is saying the leg count doesn't matter it's still an ant.
Is it just a difference in opinion about the signicance of the Overton window?
See above
Is there a technical aspect I’m missing?
Not necessarily , it's just that you are arguing a single technical issue in a conversation about perceived intentionality.
Is there some legal advantage this provides to survailance that I’ve missed?
See above
Right now, it seems like everyone is arguing against a strawman, implying that I support the idea of government/corporate surveillance and censorship, that I don’t expect that they’ll continue to be evil, or they’re simply saying its bad because its cosmetically similar to laws that do impede on freedoms. Given how unanimous the backlash is, I must be missing something?
That you are using a point nobody disagrees with to imply correctness in a context where said point doesn't really apply makes it seem like you are coming at this in bad faith.
When bad faith is assumed, people look for underlying reasons.
The fallacy is the expectation that following escalating events would arise from the event in question.
It's only a fallacy if it's unreasonable to expect the subsequent steps to occur or in this case, be attempted.
Does that mean it's a guarantee, of course not, just that the fallacy doesn't apply.
The intention or plan for escalating steps doesn't have to be laid out perfectly to draw the parallels between this and previous similar events that were then subsequently used as foundations for greater reach.
Your reasoning around the technical implementation of such escalation isn't applicable here (in the conversation about whether or not the fallacy applies)
If you want to argue that they won't escalate, or it's not possible , go right ahead, but raising a fallacy argument when it doesn't apply isn't a good start.
If you want i can address your arguments around implementation directly,as a seperate conversation? I don't think you're correct on that either, but as I said I also don't think correctness in that subject matters in the context of the fallacy.
That's on me, I meant the equivalent of a "trust me bro" , in this case an anecdotal "me and the people I know all say..."
showing Claude submissions is sufficient for analyzing code in the context I believe it is good
Yes, in the context you provided it makes sense, as a response to my question which specified examples of larger projects/workflows, it does not.
Im not here to argue either, I asked a specific question and your answer didn't really address any of it, i was just pointing that out.
I too find it frustrating but it seems for different reasons.
I really really dislike the way it's being sold as a solution for things it's in no way a solution for.
They do certain things fine, good even, but blanket statements like "their code is great" without appropriate qualifiers is contributing to the validation of these bullshit sales-oriented claims of task competency.
1: agreed
2: then I think you are missing the fundamental limitations of the current approaches, but we can agree to disagree on this.
3: see 2
I agree with jobs on the chopping block, though i think that's in large part due to poor due diligence and planing by management, but that's nothing new, the same thing has and is still happening with offshoring (throwing more people at a problem generally won't solve design and governance issues).
I also think the current systems aren't capable of being a viable replacement for anything above junior level stuff, if that ( not that that doesn't present it's own problems )
I think the difference in opinion comes from my belief that LLM's and the current tooling around them aren't fundamentally capable of replacing existing resources, not that they just don't have the power yet.
Putting increasing large compute in a calculator won't magically make it a spreadsheet application.
Not who replied to you originally but,
You aren't wrong (you even stated that more is probably better) , just not necessarily presenting the whole picture.
Ram compression isn't a benefit only scenario, there is a cost in processing power to make that happen.
So it's a trade off of memory utilisation vs processing requirements.
Whether or not it's worth it is down to circumstance, though i agree that generally i think it's worth the tradeoff.
Unified memory is useful in specific circumstances, most notably LLM/ML scenarios where high vram utilisation is part of the process.
It's not an apples to apples comparison by any means.
I appreciate it.
Yeah, I’m on the default but i'll explore the other ones now, see if there is anything i prefer.
Actually the shooting at gay night clubs in the states are usually not white Christians
citation ?
(legitimately, i've been looking for a useful resource for stats like this)
That's fair, it isn't how i read it but i can see it's importance now you've pointed it out.
This is nice in theory but the practical application is difficult.
I can get into it, but it's a common conversation i have on here, almost all of modern society is built on horrific shit, where is the line?
Is it first order disagreements like this NVIDIA boycott, or is it second order as well ? Meaning any company that willingly works with NVIDIA or explicitly buys new NVIDIA gpu's ?
It's not a trick question, I'm trying to gauge what you meant by that statement.
To be clear, i'm not saying to do nothing, i'm trying to figure out where your line is and why.
That's fair, though i would add a caveat to say "where possible"
As a concrete example, GPU's right now are ridiculous which means the secondhand market is tight, if your timeframe/need is also tight then it might not be possible to always do the "right thing™".
Agreed.
Why pick 2014, was there some regulatory requirement introduced then ?
See my question about lines above.