pksebben

19 ms·

211.
by pksebben 2y ago
Nitpick, I know, but > but with a country acting as the monopolist and therefore being exempt from antitrust enforcement. This is the US. what antitrust enforcement?
212.
by pksebben 2y ago
My guess is the system that notifies the next caretaker in the chain that someone is currently receiving CPR. if it works, there's a lot more to be done to get the patient to stable.
213.
by pksebben 2y ago
That's awful. It's such a security antipattern, obvious as soon as you take into account, you know, _Humans_... Site: Please make a password Human: 7#hs&_suiE2KcS0 Site: No copy and pasting Human: mydogisagoodboy123 Site: Nee
214.
by pksebben 2y ago
They are currently allowed to do so, and this attrition has happened without a proliferation of LLMs. This feels like an oranges-apples kinda argument. The thing that's destroying journalism is the current framework for monetizing and
215.
by pksebben 3y ago
This is a fair point. Including them is a little weird in the first place and totally a red flag.
216.
by pksebben 3y ago
Look for "diode laser cnc". If you search for "laser cutter" or "engraver" the results get mixed in with all kinds of larger machines. The world of home laser apparati is kinda wild.
217.
by pksebben 3y ago
I apologize in advance if you're sick of hearing this, but... I clicked through to your sites 'cause I dig your angle and I saw the bit about the kindle. Ouch, dude. Money sure ain't everything but holy crap. You have my co
218.
by pksebben 3y ago
The problems with prohibition are well documented. There's a good reason we re-legalized alcohol, and we suffer those same problems with the current prohibitions (the creation of a multibillion-dollar organized crime market). I think
219.
by pksebben 3y ago
Ah, I see. What I meant by that was, "A tool is as good or evil as the person wielding it". There are most definitely good and bad tools, in terms of more or less effective. Machine learning models are for sure outclassing a w
220.
by pksebben 3y ago
I'm sort of ambivalent about the death penalty, and I'm more than skeptical about punishment-based behavior mitigation in general. That said, whatever the severity of penalty you assign to direct forms of (mass) murder ought to ap
221.
by pksebben 3y ago
Okay, yeah. But even from a pretty hardcore moral relativist POV... > "mercenary spyware", AI war targeting, and death drones Are not super easy to justify. Like, sure some people obviously think those ought to be a thing, bu
222.
by pksebben 3y ago
I'm not sure I follow you - reason for what? To be clear - I'm with you that these systems can absolutely be a force for vast good (at least, I think that was what you were getting at unless there was a missing '/s'
223.
by pksebben 3y ago
Here's what I envision (note: impossible with current state of the art) A model that can be incrementally trained (this is the bit we're missing) hosted by a nonprofit, belonging to "we the people" (like wikipedia). The
224.
by pksebben 3y ago
I don't see historical dumps. Am I just dumb?
225.
by pksebben 3y ago
That gives me an idea. There are wikipedias in other languages - Maybe this framework could be adapted to translate the search terms, fetch mulitlingual sources, translate them back, and use those as comparisons. I've found a lot of st
226.
by pksebben 3y ago
There's an important difference between wikipedia and the LLMs that are actually useful today. Wikipedia is open, like completely open. GPT is not. Unless we manage to crack the distributed training / incremental improvement barri
227.
by pksebben 3y ago
On the one hand, a tool is as good or bad as the person wielding it. Smart folks with the right intentions will certainly be able to use this stuff to increase the rate and quality of their output (because they're smart, so they&#x2
228.
by pksebben 3y ago
That's gonna be a lot of fun to play with in a year or so. There's a concurrent explosion of 'veracity' analysis - it'll be fun to run those against wikipedia a year from now and your data. Incidentally, are you int
229.
by pksebben 3y ago
This method has actually been proven effective at increasing reliability / decreasing hallucinations [1] 1 - https://arxiv.org/abs/2402.05120
230.
by pksebben 3y ago
Just that the site owner, most likely, did this kind of on purpose. It's fairly unlikely that he's concerned about his "data" being used because it's junk data.
231.
by pksebben 3y ago
didn't click through to the site, didja?
232.
by pksebben 3y ago
I don't think this message is about "protecting the site's data" quite so much as "hey guys, you're wasting a ton of time and network connect to make your model worse. Might wanna do something 'bout that&
233.
by pksebben 3y ago
Between the Metaverse, "mercenary spyware", AI war targeting, and death drones, I keep wondering who it is that read Neuromancer and thought; "What a rosy picture! How can we realize this stunning vision of a future-to-be?
234.
by pksebben 3y ago
This is an incredible product; cheers to you folks. You're gonna clean up with this thing quite nicely. Light suggestion (if the creators are present); a reverse-inference function akin to Stable Diffusion's "interrogate&quo
235.
by pksebben 3y ago
We'd have to orchestrate an effort to compile the dataset. Have people listen to their favorite songs, then play dashboard drums and hum about it. Record these tagged with the actual song. You'd need a lot, though. Unless you co
236.
by pksebben 3y ago
So long as there's something to miss about human-generated content, there will be a market for that content. Things are going to get truly weird when you can no longer tell the difference, on any level.
237.
by pksebben 3y ago
Neither of those look like they have a generative AI component. We (as a society) desperately need a way to train these models in a federated, distributed manner. I would be more than happy to commit some of my own compute to training open
238.
by pksebben 3y ago
You hit the nail on the head, there. Facebook et al represented a large enough timesink / extreme enough "attention shelf" adjustment that removing them provided a dramatic and measurable change to what I was capable of. Jus
239.
by pksebben 3y ago
Right, but that's part of the problem - just being able to build more houses isn't going to solve problems for, say, NY or LA where it's not an issue of how many units but an issue of pricing
240.
by pksebben 3y ago
I don't think this is true - for one, housing in the US isn't all that scarce (~14M vacant units) [1], and housing doesn't behave like most markets vis a vis supply / demand - there are too many confounding factors and i

More ›