Ppft. Simples. We already solved this Down Under.

https://interestingengineering.com/science/biological-computer-with-human-neurons-play-doom

I have no mouth. Yet, I must scream.

World models aren’t just for robotics (though they definitely WILL be used for that). They’re for reasoning under uncertainty in domains where you can’t see the outcome in advance. Eg:

Medical diagnosis: you can’t physically “embody” whether a treatment will work. But a system that understands disease progression, drug interactions, and physiological constraints (not by pattern-matching text, but by learning causal structure) - well, that’s fundamentally different from an LLM hallucinating plausible-sounding symptoms.

Financial modeling, engineering simulations, climate prediction…all domains where the “embodied experience” is simulation, not physical interaction. You learn how the world actually works by understanding constraint and causality, not by predicting the next token in a Bloomberg article.

The point isn’t “robots will finally work.” The point is: understanding causality is cheaper in the long run and more reliable than memorizing correlations. Embodiment is just the training signal that forces you to learn causality instead of surface patterns.

My read is that LeCun’s betting that a system trained to predict abstract state transitions in any domain (be that medical, financial, physical) will generalize better / hallucinate less than one trained to predict text.

Whether that’s true? Fucked if I know - that’s why it’s (literally) the billion-dollar question. If he cracks it…it’s big.

But “it won’t cook dinner” misses the point (and besides which, it might actually cook dinner and change lightbulbs, so…)

Different approach, yeah. JEPA learns world models instead of predicting text. Whether that closes the gap with how biology actually works…that’s what he’s spending the billion to find out.

If he’d just forgone that last paragraph…

Yep. And per the article’s conclusion -

“…The question is whether being right about the problem is the same as being right about the solution.”

Tell it to Lecun. He won the Turing prize. I figure he knows what he’s doing. Let him cook I sez.

PS: I didn’t down vote you. It’s good to be skeptical.

As I mentioned elsewhere (below) I am currently conducting similar testing across 4 different 4B models (Qwen3-4B Hivemind, Qwen3-4B-2507-Instruct, Phi-4-mini, Granite-4-3B-micro), using both grounded and ungrounded conditions. Aiming for 10,000 runs, currently at 3,500.

Not to count chickens before they hatch - but at ctx 8192, hallucination flags in the grounded condition are trending toward near-zero across the models tested (so far). If that holds across the full campaign, useful to know. If it doesn’t hold, also useful to know.

I have an idea for how to make grounded state even more useful. Again, chickens not hatched blah blah. I’ll share what I find here if there’s interest. I’m intending to submit the whole shooting match for peer review (TMLR or JMLR) and put it on arXiv for others to poke at.

I realize this is peak “fine, I’ll do it myself” energy after getting sick of ChatGPT’s bullshit, but I got sick of ChatGPT’s bullshit and wanted to try something to ameliorate it.

I dunno. Some strange relic from the 1980s?

Kidding aside, it’s shocking how bad raw YT is. We watch it via Smartube (or PipePipe as needed). I can’t believe people watch “raw” youtube…it’s unwatchable.

If they ever quash SmartTube and PipePipe…well…I imagine Peertube, Nebula, Libby and Curiosity stream will suddenly become a great deal more popular.

I don’t think the powers that be fully grasp the (very delicate) knife edge they walk. They only stay in business so long as they aren’t annoying enough to be replaced. Actually, who am I kidding - they know that exactly and play the delicate “gently gently” boil frogs in a pot game like grandmasters.

Yep, same issue with Firestick here.

What I hear you saying is you have great taste in consoles, a kick ass TV and a free space heater.

It seems like a lot of the bigger names suck. I bought a Blauerpunkt and it is awful - not hackability wise but as a product. Probably for the same reasons as Nokia, Phillips, JVC etc are pale shadows of themselves (sold off / rebagged)

I have a Blauerpunkt, a TCL and a Samsung. Of the three, it’s the TCL that’s been the least locked down.

At this rate, I’m probably going to go for a short throw projector or just get an old school plasma if /when these go tits up.

Well, I was mostly piggy backing on your comment / signal boosting what you were saying for the cheap seats. I figure anyone who’s hacking their TV via ADB commands doesn’t need my help with work arounds.

But heck, if any of those are of use, please enjoy.

I’m tempted to mentioned a good “yarr me harties” source here too, but I don’t want to teach anyone how to suck lemons.

True… but that was greatly mitigated in later models.

I swear that the motion quality and the deep blacks rival high end OLEDs. Really, really good. So long as you can find a decent TV (like the one in the video).

The issue I have with them is - power draw and heat.

Well, in fairness, Flauncher and Nova launcher are on the play store. Irrespective, there are ways to install things now and then harden against google, such as blocking MAC address, Pinhole, disabling updates etc.

Alternatively, an android dongle (like Onn) is an option, however I would advise against Amazon Fire sticks and the like (telemetry, locked down)

Alternatively, make your own. A Raspberry Pi CM4 TV Stick that plugs into HDMI is one known option. If Andorid (and ASOP) become as locked down as you fear, there will be other options - Kodi, LineageOS TV etc.

If Android wants to shit the bed, no problem. It’s not the only game in town.

Would you consider an old plasma TV? They won’t be in 4K (I think they topped out at 1440p) but there are some real advantages to them (if you can tolerate the disadvantages).

https://www.youtube.com/watch?v=jIuKA2DUgas

Good call!

I’d also like to share another option for folks: Flauncher. I’ve been using it for a few years - it’s very clean and lightweight.

You can set it as the default using ADB commands, or install it directly from the Play Store and use adhoc.

https://gitlab.com/flauncher/flauncher

Paired with a button remapper and set as default, I honestly haven’t seen the stock Android interface in years. (Like you, I’ve also used ADB to disable telemetry etc on TV, and replaced google play with Fdroid and Aurora)

PS: For the uninitiated, these are also worth installing:

https://smarttubeapp.github.io/

Gives you ad-free YouTube and then some.

https://f-droid.org/packages/org.courville.nova/

Lets you connect a hard drive to your router and create your own local media streaming frontend ala Netflix (or plug your hard drive / USB stick directly into your TV; it doesn’t care).

While it lacks some of the polish, it’s simpler than Jellyfin or Plex.

Sorry…can you explain that a different way? I’m not following what you’re trying to say.

YTMND

I understood that reference.

We’re doing a great job of that ourselves, so mission accomplished?

George Dubya, is that you?

I’m also skeptical as to the FV ever matching Reddit in terms of variety and bulk of content. The situation just is what it is.

…from your fingers to God’s eyeballs.

That seems like… a little much. I do agree that upvotes/downvotes indeed gamify the system, but on the whole would say that the end-effect on Reddit results in a big bunch of hoomons acting in typical hoomon ways, which is with deep undercurrents of fickle, ignorant, selfish, feel-good behavior.

I’d argue that’s a restatement of my position with better adjectives :)

I’d say I agree with most of the things you wrote, but remain unconvinced that upvote/downvote is so absolutely toxic as to merit tossing. And of course, I don’t think it’s going to happen, anyway.

Aggregate behaviour amongst naked apes? Yeah, I would tend to agree. Now what?

Well, 2 options:

1. Kill all the apes (or just wait 15 more minutes)

2. Enjoy Lemmy

I’m trending towards 2 myself