FaceDeer

The word “slop” has lost all meaning.

If you’d like to analogize with special-purpose programs versus general purpose ones, you could consider an LLM to be more like an “operating system” rather than a single application running on that system. The LLM isn’t specialized in a particular task, it supports running any task you want to throw at it.

In this analogy, the specialized applications that are running on the LLM would be the things called “agents” and “harnesses.” They’re the parts that hold the code that is tailored specifically to the particular task that they’re for. So if I wanted to set up a system using an LLM to, for example, read court transcripts and automatically search legal databases for statutes and case law relevant to whatever’s being discussed in them, I wouldn’t simply copy and paste the transcript into the LLM’s context and expect something useful to come back. I’d have to embed the LLM inside a system that prompts it correctly, has tools that can search legal databases, mechanisms for storing intermediate results, scripts to check the formatting of inputs and outputs, and so forth. None of that accessory stuff needs to be an ML model, it can just be conventional programming. Trying to train a specialized ML model from the ground up to do all that stuff without the associated harness helping it would be hugely wasteful.

I suppose another analogy could be the hardware itself. For every task that we use general-purpose von Neumann architecture CPUs for we could create a specialized chip on a purpose-specific circuit board. But instead for most tasks we find it’s much cheaper and more convenient to create general-purpose computers and then program them for these specific tasks. Nowadays you’ll often find that simple home appliances with just a few buttons and a few functions will have a full blown microcontroller inside them with firmware. Probably lots of unused inputs/outputs and ram and whatnot. That’s because it’s far cheaper and easier to build a factory that stamps out millions of general-purpose programmable chips than it would be to have hundreds of factory lines that each do a run of ten thousand custom-designed chips. It’s genuinely less wasteful doing it that way.

His most recent official line.

The next thing out of his mouth could be completely different and that will become the official line at that point. He only sticks to any sort of agreement or position for as long as it benefits him, not one second longer than that.

Trump hasn’t backed down from the 51st state thing permanently, he just gets distracted. Trump is deep into dementia and one of the things that results in is obsessions that simply cannot be let go of. The 51st state thing will keep coming back, just like the Greenland thing.

What stops him is that even though he managed to ensure that there are no adults in the room this time around to rein him in like last time he was in office, even the children in the room that he’s surrounded himself with realize that this is a terrible idea.

I would hope so, but you’re laying it on pretty thick with this hypothetical scenario. Makes it hard to imagine.

Success for Rutte (and NATO) is protecting NATO members against invasion or attack.

If the US’ membership is raising that risk for the other members rather than lowering it, then in that situation letting the US exit can be a net benefit for NATO’s members (which would no longer include the US).

The point is to make belligerents back down by raising the cost they’d have to pay by carrying through.

This particular belligerent backs down so often that there’s a special acronym for it, so NATO’s a perfectly good protective tool here.

40 years is an unreasonably long timeframe to be making projections like this. Assuming the US is still having elections there will have been 10 opportunities for their government to completely flip-flop its intentions in that time.

There have recently been direct annexation threats against Canada.

Yeah. It’s no longer okay to have a plan that says it’s okay if Russia takes some territory, we’ll just take it back a little later. We’ve now seen what happens to the people living in territory that happens to.

And even if small places could survive an invasion attempt by Russia, it would still have an immense cost in lives and destruction. Why not prevent that in the first place?

Or, alternately, preparing NATO for America’s exit. I could easily see a scenario where they’re more trouble than they’re worth.

I didn’t say anything about the “prevent instances from being overloaded” part being good or bad. I didn’t even give an opinion on ActivityPub, just pointed out the practical limitations and incompatible design goals.

Personally, I’ve got no problem with websites implementing rate caps and whatnot to ensure that their traffic remains within the limits they can handle, or throttling specific IPs. I am very concerned with how Cloudflare in particular has become the single centralized “gatekeeper” for vast swaths of the Internet, though. If they decide that some particular client isn’t allowed to see stuff then poof, a big chunk of the Internet is cut off. That’s worrisome IMO.

Your original post didn’t specify a particular kind of data scraping. TropicalDingdong had no way to know you were only specifically interested in that one kind of data scraping, so his comment is appropriate - you can’t stop data scraping in general, and attempting to do so in the general case goes directly against the goal of ActivityPub.

This sort of nonsense, along with idiocy like that guy who uses the þ character in place of “th” in all his comments, does absolutely nothing to hinder AI training. The only effect it has is to bother human readers. Completely counterproductive.

That’s a very narrow view of data scraping, there’s lots of ways to get data.

The Fediverse is built on ActivityPub, which is an open protocol that’s designed to broadcast data with no limitations or restrictions. If you don’t want your data to end up in the hands of anyone who wants it - including those nefarious AI trainers - then that’s an inherently incompatible goal with ActivityPub.

If you’re just worried about specific instances being overloaded with requests, then sure, all the usual rate limiting DDOS-prevention Cloudflare tricks will work. But the data itself isn’t “protected.” Someone who wants it could simply run an instance of their own specifically to collect it.

Indeed! When I saw that title I sighed and thought “alright, time to post the link to the Dartmouth workshop where the term AI was coined in 1956 and get downvoted a bunch for it.” But someone else took my bullet? How dare!

And worse, the thread in general seems to be filled with mostly rational comments about nomenclature and stuff!

I don’t know how to respond to this.

It’s so dumb that these things are done “in rebuke” to things.

Israel should recognize the Armenian genocide because it fucking happened. As should every other country, for whatever it’s worth that a country officially “recognizes” facts like these.

Could just add it to their demands for opening the Strait of Hormuz.

He was an incredibly prolific poster. Unfortunately I found him to be rather abrasive, I don’t recall any particular interaction but I do get an “oh yeah, that guy. Glad he’s not been around much lately” sense when I see that name.