"Here's my rule: I should be allowed to do whatever I want and it's your job to deal with it without criticizing my shitty behavior"

no wonder you're running defense for the dem's dogshit political instincts, you're a conservative and you like it when they do loser shit lmao

But they don't list it as a demand. they list it as a show of good faith. Are they willing to give ICE *more* money without removing the alleged source of of the problem?

because even if you *were* correct and they were specifically demanding the rescinsion of the memo, then their demands are also still weak and ceding ground because it does not demand immediate removal of leaders attempting to actively circumvent people's constitutional rights

Answer me this: if they believe the memo is the main problem here, why do they cite removing Noem would be a show of good faith rather than demanding she be removed and investigated/prosecuted for directly and knowingly directing her agency to violate the constitution?

learn. to. read. "we believe congress needs to enact the following guardrails" implies these guardrails do not already exist, it's arguing from a point of weakness, and that's ignoring that accepting literally any compromise at all on any of these demands weakens your constitutional rights

I am not reacting to a headline, you're missing my point, wokr on your reading comprehension.

problem is that doesn't even begin to address a single 4th amendment concern, but by even pretending to negotiate by making demands for a new policy to follow the 4th amendment, dems are implicitly accepting stupid excuses like that one as somewhat valid

If somebody broke into your house, said they lived there now, and then you started to negotiate over which room you get, you've already ceded ground.

negotiating for ice to follow the fourth amendment implies that they currently do not have to--they're accepting the right's framing on the issue by accepting that warrantless entry is the status quo that needs to change

fun fact, there are some specific circumstances under which you literally do in fact have to hand it to ted kaczynski:

and considering you seem to be more of a debate pervert than someone actually interested in the topic and I have a family that actually loves me, im gonna do us both a favor and make it so neither of us has the displeasure of talking to each other anymore :)

yeah, I was talking about it being useful for actual practical purposes, the actual criticism you dont seem to want to engage

you're pretty good at gish galloping, I'll give you that

putting words into my mouth quite a bit there for someone pretending to be coming to this in good faith, bucko

nah, I dont really care too much about this line of reasoning--we're both extrapolating, I'm extrapolating more conservatively because most engineering problems of this sort tend to run into diminishing returns, and that's all there really is to it. future prediction is messy.

It's not a misunderstanding on my part, and I dont really care what you'd rather focus on when you're trying to steer away from the main problem with LLMs: they aren't useful

when you're measuring automation of any kind, even odds is horrifyingly bad

automation below the 80% mark would be unacceptable in literally any production process

this is not an average performance measurement though--it's an estimate of how long it takes a human to do a task that the ai is a coin flip for. that's a different measurement than you're suggesting.

do ypu know how machine learning and training works? because I'm not gonna give you a class on it to win an argument. there are diminishing returns on training data, and you can either read about it on your own time or take my word for it i dont care which.

this is not measuring mean times between failures, and they have released clarifying statements about this explicitly because that is *not* what they are claiming.

diminishing returns on training data and power usage. there's a limit to how good these things can even theoretically get, and I'm willing to bet that we're much closer to that limit than you might think.

(the 90%+ range is the only benchmark where I'd start to agree these might have *some* utility)

that is, of course, assuming this contrived 50% success horizon is actually a useful metric to begin with, which by my lights stretches credulity more than the initial claim in the first place, *especially* when they note that this looks absolutely abysmal at higher success rates,

which is a completely unfounded assumption--you cant just linear regression your way into the future on this kind of thing that you have very little data on, especially when the most likely scenario is a plateau in improvements given the diminishing returns on training in the last year or so

I'm not staying up much later to argue with ai boosters so I'll just stick with the first one for now: There is no compelling reason to accept their metric as a useful baseline at all. additionally, they are arbitrarily assuming that they will continue to improve at similar rates,

you're literally making the "if NFTs are worthless why are people buying them" argument! this is literally every tech fad of the last couple decades on repeat!

there are not continual measurable gains! you're actively delusional!

i have, and I've also asked the questions any rational person would:
who's supplying the benchmarks
what do the benchmarks *actually* measure
are those good indications of the claim being made
do those benchmarks refute any prior data?

you, clearly, have not asked those questions

my point is your anecdotal experiences are not relevant.

that study that showed Ai use slowed programmers down? it also showed that those same programmers overwhelmingly *believed* it was helping them. there is no reason to believe your experience is any different.

"it'll get better because we can train it to be better! nevermind the fact that marginal improvements are taking exponentially more training data, and pay no attention to the amount of new quality training data there is left to train on! Tech always gets better no matter what!"