Hacker Newsnew | past | comments | ask | show | jobs | submit | abraxas's commentslogin

Great, this is 'Manna' portrayed future coming to us.

Before the dystopian black and gray fad arrived most buildings that went up were sort of OK. And I didn't mind the pastel paint on commie blocks either. But a decade ago someone decided that gray cuboids with asymmetric windows were an improvement...

Even so, I think North American cities are on average uglier than most Polish ones. Overall we're not doing so bad but I want the Slavic city memes to continue lest we get Prague or Amsterdam level tourist invasion.


Excellent idea. So many modern buildings age so poorly. Maybe this will give some starchitecs a bit of a pause...

Doubt it. Demoralization is what they're after.

If there is one future tech I hope we never reach it's this. The downside risks are even worse than runaway super intelligence.

why's that?

Infinite life, finite resources, comes immediately to mind.

Not aging does not mean infinite life. Not only can you end it whenever you want, but you can still have accidents or win a Darwin Award.

Also, longer lifespans open up space exploration and so much more. Anyway, I don't know if I want infinite life, but I definitely want us to beat aging, it sucks from every possible point.


The people that will be accessing this technology are not the people you want to have access to this technology.

What about Fei Fei Li's lab? I think they are generating true 3D worlds rather than frames of a video?

Although that probably precludes her from having animations in those worlds...


> if you wanted imagination you don't need to make a video model. You probably don't need to decode the latents at all.

Soft disagree. What is the purpose of that imagination if not to map it to actual real world outfcomes. For this to compare them to the real world and possibly backpropagate through them you'll need video frames.


You elected yourself oligarchy. Good luck getting rid of it now.

Yes, absolutely yes, The dotcom bubble was precipitated by lots of people in mainstream media sounding the alarm. Ditto for the real estate bubble that led to the stock market collapse of 2008.


I personally remember the string of “us re makes no damn sense” articles in the financial press end of ‘06 and early ‘07. Iirc credit ratings and mortgage exposure specifically was being widely panned by 07. And still it took another year for it to shake out to the financial system at large.


It seems that something that does away with human friendly syntax and leans more towards a pure AST representation would be even better? Basically a Lisp but with very strict typing might do the trick. And most LLMs are probably trained on lots of Lisps already.


Generally seems a bad idea to have your LLM write languages you do not understand or write yourself


Doesn’t that apply to the OP as well?


Yes, I'm not going to fill my precious context with documentation for a programming language

This seems like a research dead end to me, the fundamentals are not there


It seems kind of silly that you can’t teach an LLM new tricks though, doesn’t it? This doesn’t sound like an intrinsic limitation and more an artifact of how we produce model weights today.


getting tricks embedded into the weights is expensive, it doesn't happen in a single pass

they's why we teach them new tricks on the fly (in-context learning) with instruction files


Right, it sounds like an artificial limitation.


it's more a mathematical / algorithmic limitation


I’ll counter it’s an architectural issue


I would put that under the umbrella of algo/math, i.e. the structure of the LLM is part of the algo, which is itself governed by math

For example, DeepSeek has done some interesting things with attention, via changes to the structures / algos, but all this is still optimized by gradient descent, which is why models do not learn facts and such from a single pass. It takes many to refine the weights that go into the math formulas


> I would put that under the umbrella of algo/math, i.e. the structure of the LLM is part of the algo, which is itself governed by math

Yes you’re right. I misspoke.

I’m curious if there are ways to get around the monolithic nature of today’s models. There have to be architectures where a generalized model can coordinate specialized models which are cheaper to train, for example. E.g calling into a tool which is actually another model. Pre-LLM this was called boosting or “ensemble of experts” (I’m sure I’m butchering some nuance there).


> The apocalypse is delayed, permanently

That is your survivorship bias. There are societies that collapsed never to rebuild or became mere shadows of themselves. I'm not just talking complete collapses like the Easter Island or the Mayan civilization. In very recent past we witnessed the cultural collapse of Japan after their bubbles of the 1980s burst and derailed their economy. But the most spectacular collapse has been that of the Soviet Union. The moral and cultural sickness of that society is now on full display. It's a potemkin village of a society. On surface things look like they still hang together - they still have electicity, internet and even their iCrap. But the will to live, the idea that there is a better tomorrow is all gone from that society. It's a nation of ghosts who don't live but merely exist in a post collapse stupor. Things still function just enough to not spark a revolution but almost nobody finds it fulfilling to live.


I would argue you're just not zooming out far enough. Those things were definitely a society-wide collapse, and horrible for the people who lived through them, but humanity as a whole has just continued to grow, develop, and thrive, at least from my perspective.

Let me put it this way: is there a time anywhere in the past 300,000 years where you'd prefer to jump back to if you were going to be born as a random person anywhere on earth?

In spite of everything we face today, I struggle to think I'd want to be born anytime in the past.


> There are societies that collapsed never to rebuild or became mere shadows of themselves.

Yeah but the people were still there. Collapse of a society is a change in people's point of view, not always (or just) fiery death for everyone. Armageddon is a change in social order.

To bring this closer to home; the dot-com bust wasn't really noticed by cash positive startups with clients in the real world. One person's collapse was another's Herman Miller and Ducati Monster sale.


You're talking about the "cultural collapse" of both Japan and Russia as if it was common knowledge. What exactly do you mean by this? Is this your personal opinion, or a reference to some quantifiable metric?

Japan is currently one of the hottest tourist destinations in the world. First because of the strength of the dollar vs the yen, but also because of their culture.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: