A lot of people don’t understand how AI training and AI inference work, they are two completely separate processes.

Yes, they are. Not sure why you are bringing that up.

For those wondering what the actual difference is (possibly because they don’t seem to know):

At a high level, training is when you ingest data to create a model based on characteristics of that data. Inference is when you then apply a model to (preferably new) data. So think of training as “teaching” a model what a cat is, and inference as having that model scan through images for cats.

And a huge part of making a good model is providing good data. That is, generally speaking, done by labeling things ahead of time. Back in the day it was paying people to take an amazon survey where they said “hot dog or no hot dog”. These days… it is “anti-bot” technology that gets that for free (think about WHY every single website cares what is a fire hydrant or a bicycle…)

But that is ALSO just simple metrics like “Did the user use what we suggested”. Instead of saying “not hot dog” it is “good reply” or “no reply” or “still read email” or “ignored email” and so forth.

And once you know what your pain points are with TOTALLY anonymized user data, you can then “reproduce” said user data to add to your training set. Which is the kind of bullshit facebook, allegedly, has done for years where they’ll GLADLY delete your data if you request it… but not that picture of you at the McDonald’s down the street because that belongs to Ronjon Buck who worked there one summer. But they’ll gladly anonymize your user data so the picture of you actually just corresponds to “User 25156161616” that happens to be the sibling of your sister and so forth…

in fact a lot of research is being done right now trying to make it possible to do both because it would be really handy to be able to do them together and it can’t really be done like that yet.

That is literally just a feedback loop and is core to pretty much any “agentic” network/graph.

Go ahead and do so, they will have separate sections specifically about the use of data for training. Data privacy is regulated by a lot of laws, even in the United States, and corporate users are extremely picky about that sort of stuff.

There also tend to be laws about opting in and forced EULA agreements. It is almost like the megacorps have acknowledged that they’ll just do whatever and MAYBE pay a fee after they have made so much more money already.

Understand that basically ANYTHING that “uses AI” is using you for training data.

At its simplest, it is the old fashioned A/B testing where you are used as part of a reinforcement/labeling pipeline. Sometimes it gets considerably more bullshit as your very queries and what would make you make them are used to “give you a better experience” and so forth.

And if you read any of the EULAs (for the stuff that google opted users into…) you’ll see verbiage along those lines.

Of course, the reality is that google is going to train off our data regardless. But that is why it is a good idea to decouple your life from google as much as possible. It takes a long ass time but… no better time than today.

As it stands? Cloudflare is still incredibly effective at protecting customers from those DDOS attacks. Which, depending on your hosting solution, can mean very noticeable monetary savings because YOUR hardware/connection didn’t spike. And, regardless, can mean noticeable monetary savings as your engineers didn’t need to recover a crashed system because your setup was just sitting there idle.

That said: If you truly need high availability? You need to do what downdetector did and have alternatives ready in the event that Cloudflare falls over. Same as with your ISP… which should be ISPs plural.

deleted by creator

1. Poe’s law?
2. Endlessly sucking off cheney and mao in hell?

Homie. You’re asking for help. Provide the info up front rather than expect people to dig through your entire mess to collect basic details.

Other useful information: Is this raw dog or flapak steam? At a glance that looks like a modal controller. What mode(s) have you tried connecting it as?

And, generally, usb dongles don’t work as well as bluetooth for gamepads in linux.

Yeah. The US definitely was horrifically evil long after other Global North countries had decided to chill but we were far from the only people running on the genocide and enslavement of the Global South.

In a lot of ways, the US was late to colonialism and took it WAY too far and commit atrocities for it. The USSR/Stalin was late to (there is probably a better term for it but I can’t remember it) peasant revolutions and took it WAY too far and commit atrocities for it… and then ended up being a global power that extended said reign of terror to anyone sharing a border with them.

Communism inherently couples both the economy and the government.

In theory, capitalism can be decoupled since it mostly depends on laissez-faire governance. Communism inherently requires a planned economy and centralized control of such.

There is theoretically nothing stopping said leaders of a communist regime from being elected through a democratic process. But much like democracies tend to favor capitalism and (lower case) libertarian ideals, communism tends to lend itself to dictatorships because… you have a centralized control of all aspects of society.

dot ml 'bout to crest the horizon like the riders of rohan.

Agentic AI is just a buzzword for letting AI do things without human supervision

No, it isn’t.

As per IBM https://www.ibm.com/think/topics/agentic-ai

Agentic AI is an artificial intelligence system that can accomplish a specific goal with limited supervision. It consists of AI agents—machine learning models that mimic human decision-making to solve problems in real time. In a multiagent system, each agent performs a specific subtask required to reach the goal and their efforts are coordinated through AI orchestration.

The key part being the last sentence.

Its the idea of moving away from a monolithic (for simplicity’s sake) LLM into one where each “AI” serves a specific purpose. So imagine a case where you have one “AI” to parse your input text and two or three other “AI” to run different models based upon what use case your request falls into. The result is MUCH smaller models (that can often be colocated on the same physical GPU or even CPU) that are specialized rather than an Everything model that can search the internet, fail at doing math, and tell you you look super sexy in that minecraft hat.

And… anyone who has ever done any software development (web or otherwise) can tell you: That is just (micro)services. Especially when so many of the “agents” aren’t actually LLMs and are just bare metal code or databases or what have you. Just like how any Senior engineer worth their salt can point out that isn’t fundamentally different than calling a package/library instead of rolling your own solution for every component.

The idea of supervision remains the same. Some orgs care about it. Others don’t. Just like some orgs care about making maintainable code and others don’t. And one of the bigger buzz words these days is “human in the loop” to specifically provide supervision/training data.

But yes, it is very much a buzzword.

Steam Frame is going to have a snapdragon processor and will “be a computer”.

So there will be an ARM native Steam client (and Steam OS that may or may not be SteamOS). Just a question on whether that gets a wider release.

But yes. Games themselves will be heavily dependent on FEX.

Related: has anyone actually tried Alyx on Linux recently? I set up ALVR with my facebook quest the other month and every game except Alyx was fine. Half-assed googling makes it sound like that was a known issue but there are also probably like five people on the planet trying Alyx in Proton.

From everything we have heard… I would be shocked if it wasn’t pretty damned close.

Gamers Nexus touched on the pricing info they were given. Go watch the video to confirm but off the top of my head:

• The Steam Machine will be priced competitively with an entry level computer
• The Steam Frame will be below the price of an Index

So what that translates to is

• The Steam Machine will likely be in the 800-1500 USD range
• The Steam Frame will be up to 1000 USD

Which… sounds about right. The Steam Frame is going to use a comparatively cheap Snapdragon processor but it still needs all the HMD tech. The Facebook Quest 3 is around 500 USD and considering economy of scale… that is probably the price floor for the Steam Frame.

And the Steam Machine? That is rocking a proper Zen 4 with 16 gigs of DDR5 and 8 gigs of DDR6. Considering how expensive RAM already is and how that probably ain’t going down until late 2026 at the earliest… And it is worth noting that people lost their shit over the ROG XBOX ALLY X S 45 WHATEVER being 1k but… spec wise that lines up with similar laptops. The display is a decent chunk of that, which the Steam Machine won’t have, but… yeah.

Computers is expensive. Especially in a Post Liberation Day world. It will be a miracle if the base console price (because you can bet the PS6 is gonna do the same stupid bullshit MS did with the Series S…) is below 900 USD with the “real” price being well over 1k. And the Steam Machine is going to be priced along those lines because Valve (presumably) doesn’t have a bunch of warehouses full of parts from five years ago.

The good news is that if you already have a gaming PC, and don’t need the Valve branding, you can get a pretty solid AMD NUC for 300-600 USD that will run Bazzite perfectly and play a lot of your games locally with the rest streaming over Moonlight or Steam Link. GMKtec pretty much have this market on lock and I personally love my K11 (overkill but also really nice to not have to walk upstairs to wake my desktop for every single game).

You’ll have the same nonsense with HDMI 2.1 as the Steam Machine will (so VRR) and AMD but there are workarounds for that (basically you flash a displayport dongle to be REAL sketchy). And you’ll be able to take advantage of most of the software improvements Valve are pushing for SteamVR, SteamOS, and Steam Link that are going to be coming rapidly for the launch. MUCH less oomph but… people who are expecting proper 4k experiences out of a Steam Machine are lying to themselves.

You would need REALLY massive hands to touch those when your thumbs are resting on the analog sticks or the face buttons.

Not sure what drugs you are on but… gimme some.

But yeah. I am glad that the world is trying to minimize dependence on MS. Nextcloud… is very much not it. It is bloated and plugins break constantly and it is shockingly horrible for everything except file sync.

There is no such thing as an “ongoing ‘clean’ data source”. Because the people paying for these models are actively using it on those platforms. Whether it is because they are too stupid to express their own thoughts and “use chatgpt for coming up with ideas” or because they are running a bot farm.

And the reality is that… it doesn’t actually matter. People are on the “dead internet theory” bandwagon again. But… think back to the past five or six years. How often have you had a meaningful conversation on ANY social media platform? It happens occasionally, but mostly you can hope for someone to key in on a single sentence you said or actively misrepresent your post because they wanted to tell a joke they clearly have been workshopping for the past few days. More often you get someone who just keys in on a single word and pastes a copypasta that they know will do well (see: Basically any topic about “AI” on the fediverse…). And that is assuming you get a reply at all.

Let alone all the people who pretend the internet used to be a wonderful place where you get the answer to everything instantly rather than worrying if chatgpt hallucinated. Those motha fuckers clearly never spent much time on stack overflow or the intel message boards…

“AI” didn’t learn how to pass the Voight-Kampff test. Humanity now does so poorly on it that we have to grade on a curve.

So as for “sloppy” models? We are probably less than a year out before influencers start bragging that they love ShitGPT 7 because “it isn’t pretentious. It talks like me”. Sorry “Others disingenuous. ShitGPT speak truth gooder. Like, comment, subscribe and use my affilly linkle”

Like… OnlyFans ostensibly isn’t a porn platform. But they know what they’re doing. The credit card companies know what they’re doing yet turn a blind eye? How does that work?

OnlyFans corporate actively downplays the porn and STILL try to pretend it is about normal vlogging and fitness. It is a big chunk of why there was that former disney channel actress who made a huge deal about posting some bikini shots to OF. She made bank and OF got themselves in the news as “actually it is all not even softcore”. Similarly, OF has shockingly strict and well defined rules for the kinds of content that can be sold (stuff like “no fisting”).

And the payment providers generally don’t care until they are legally forced to care.

Same with Reddit. Karma farming is horrible for the site’s long term health, yet they seem to structure things to encourage it as much as they can.

Long term doesn’t matter but also…

Reddit’s product isn’t people pretending they are clever for referencing an eighteen year old joke from It’s Always Sunny or having endless arguments over whether someone is morally virtuous for posting a meme about someone they don’t like being a bad pet owner.

The product is the advertisement and “authority”. Having someone respond to “what wok should I buy” with a summary of why you kind of want the cheapest carbon steel piece of crap you can find with the desired diameter, bottom, and handles isn’t helpful. Blog sites already exist. Hell, you can check the heat map on the youtube search bar for how many people just skip to the end of a Project Farm video.

But by having the top result be whatever brand won that week’s fight? Suddenly everyone talks about how much they hate google and how the real secret is to search “best wok reddit”. And then if reddit suddenly starts charging for data scraping so google can’t see it anymore or API access to automate those kinds of “grass roots” answers?

As for long term? There is no long term. They have years of training data. They’ll get bought out by whoever wants it the most and then be done with it.

I kinda have a distant fascination with this whole ecosystem. Like, what drives all this money and attention to change hands? I don’t really see the appeal of idolization, myself, which is why I find it interesting I suppose.

Only Fans? People be horny and the most successful sex workers understand that it isn’t the stripper who shows her poon on stage that makes the most money: it is the one who can work a crowd and manufacture a “connection” with the guy who will pay for five lap dances a week. Its why so many of them will sell private videos and text messaging as an incentive (and then, when they make enough money, pay a service to do said texting). At which point it is the old screenshot of a kid commenting on a pewdiepie video that they are going to mow the lawn and then they are back… except it is someone typing one handed to a porn vod.

Advertisement and astroturfing? Advertisement works. People pretend they are above it but they aren’t. One of my genuine favorite examples was back during the pandemic when one of the WWE shows came on right before some reality tv show called Sex Island or whatever. You could watch in real time as people began by complaining they hate all the ads and it is trashy (it really lowered the vibes of the sex trafficker company…). But, after a few weeks, there were “ironic” watch alongs on all the major wrestling message boards.

And that is especially true on reddit and the like. There is a reason that “SEARCH STRING reddit” was a “lifehack”. And people are generally smart enough to call out the account with -20 karma that is talking about how awesome the Babish Wok is. They are less likely to do that when the account has 5k karma and the last few comments look legit.

And… if you know the right circles to look in, there are still orgs that will pay for “real” social media accounts either for one offs or for ownership.

Sometimes it is a hardcore simp who has keyed in on the complete lack of discoverability in the OF space.

More often… people are just deeply fucking stupid. Its the same reason you’ll see people incorporate the same obnoxious self-censoring that tiktoks do in their posts… because they know that pewdiepie is afraid of getting demonetized for saying “pedophile” so they might be too. Or… the dumbfucks who shout into a speaker phone in public because they saw kim kardashian do it on the tv.

And same here. They are keyed in enough to realize that sites like reddit are full of people who post porn to get their accounts enough karma to get past most filters without realizing that the goal of that is to then nuke the past messages and sell it to a marketing firm (which is why reddit finally added the ability to hide an account’s posts and comments).

I mean… the good news is that sex workers clearly think it is worth advertising here.

Except… probably not. A good friend of mine has her content posted somewhat prominently here and on mastodon. She herself is not involved in the slightest and it is just someone karma farming in systems without global (or local) karma.