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

Let's say an agent needs to do 10 brain surgeries on a human to remove a tumor and a human doctor can do it in a single surgery. I would prefer the human.

"steps" are important to optimize if they have negative externalities.


This really seems like a nothing burger.

The $1B were a refund. Net exchange ~$0.

Building out fossile fuel production shifts oil revenue from various dictatorships around the world to the US in this case. That's a good thing. I wish we in Europe produced more gas ourselves instead of being highly dependent on other countries.

This does not mean higher gas demand, which is what matters for CO2 reduction.


I would argue it’s a waste of resources. Fossil fuels will be phased out over the next 50 years. If China is an indicator it can be done quite a bit faster than you’d think.

It’s already questionable to build fossil fuel capacity at today’s prices. In 10 years it won’t make any financial sense. In 20 you’d be laughed out of the room.

Why waste money on a dying technology? It’d be like mining bitcoin


(Based on AI research) fossile fuel production takes a few years to recoup investment. Only refineries seem quite long term (15-20 years). I'm sure the companies have projections about future demand to decide which projects are worth pursuing. If they miscalculate, the oil companies loose money (oh no! /s). Anyway, you won't loose money on this.

There is of course also an argument about national security, not being at the whim of some Iranian dictator. So some form of government investment would be justified, but not necessary IMO.


I do lose money on this. I (and you, and the rest of the world) have to both clean up emissions and pay for climate change. The more we emit the worse the effects.


Right, but that negative externality is on emissions. For emissions it doesn't really matter who is producing the oil.


Or we could invest in renewables. Renewables get cheaper than oil when you do that. Then nobody produces oil.

The production shifts oil revenue from Islamic dictatorships to Christian dictatorship of USA. Maybe a hyperbole for now, but for how long? Independent media and education under attack, gerrymandering your voting rights away etc. Considering most of the population lives is solid red or blue state and thus their vote for president doesn't count, how much of a democracy is USA when the minority of independents live in swing states are the only who's vote counts?

Also USA us doing nothing relevant to reduce gas demand like CO2 cap and trade or CO2 based tariffs.

At this rate buy Chinese will be a more moral choice than buy American by next decade.


Voters in the US have the ability to actually change the course of their country. Voters can say they want Obama or Trump in charge and the government has to obey that choice.

In Europe you only have the same parties/"uniparty" in power all the time. Many people never had a representative they voted for.

I understand the nuance about the US voting system. But when I look at the outcomes, the US seems way more democratic than Europe.

Re gas demand: Renewables are cheaper than fossile fuels. They will obviously win out in a free market. No subsidies needed.


They only had to comply with EU laws when they were already a big player in China. EU manufacturers need their new vehicles to be compliant on day one. That is, if they want to launch in the EU market first. Audi recently launched a China-only car (AUDI E5 Sportback).


What is different about Deepseek's use of MoE vs all the other MoE models that makes training more efficient?

FP8 training and GRPO make sense to me, but that only gets you a 4x improvement total, right?


They slightly restructure their MoE [1], but I think the main difference is that other big models (e.g Llama 504B) are dense and have higher FLOP requirements. MoE should represent a ~5x improvement. FP8 should be about a ~2x improvement.

We don’t know how much of a speed improvement GRPO represents. They didn’t say how many GPU hours went into to RLing DeepSeek-r1 and we don’t have a o1 numbers to compare.

There’s definitely lots of misinformation spreading though. The $5.5m number refers to Deepseek-v3, not Deepseek-r1. I don't want to take away from HighFlyer's accomplishment, though. I think a lot of these innovations were forced to work around H800 networking limitations, and it's impressive what they've done.

[1] https://arxiv.org/abs/2401.06066


It's interesting that only having access to less powerful hardware motivated/necessitated more efficient training--like how tariffs can backfire if left in place too long.


If you don't, your geopolitical adversary might be the first to build AGI.

So in this scenario I could see it become necessary from a military perspective.


The mere existence of regulation is part of the problem. Without precise understanding of the law, you don't know if your use cases are fine/excempted. The safe default assumption is that your site is not compliant with regulations until you can prove otherwise, involving a lawyer.


That regulation would not have come into existence if there were no privacy problems caused by the ones that have to comply to the regulations


Right, and that regulation has a cost. I hope it's worth it.


There were certainly problems.

But the GDPR and ePrivacy directives don't protect us from nefarious cross-site tracking cookies.

Prior to the GDPR, websites just tracked us.

Now they track us AND present an irritating warning that users have learnt to mindlessly "accept"


This is simply not true.


Once you have a technological breakthrough that requires lots of exploration to figure out which products will succeed, regulation is a competitive disadvantage.

For established industries regulation just increases prices. ... Until a new technology comes up that allows new competitors to enter the market. Like Tesla, SpaceX.


Hmm both SpaceX and Tesla are heavily subsidised by the government

Funnily enough many of the basic breakthroughs in AI were done in Europe, but at universities.

Sadly people don‘t invest in Europe in startups as in the US, therefore the companies were formed in the US, even tough research is done here.


If you declare using an invalidated iterator as UB, the compiler can optimize as if the container was effectively immutable during the loop.


Ignoring it means you need to get explicit consent, which is what the websites are already doing.


Legally they are required as of now to gather informed consent that is given freely.

Contrary to popular believe the EU has somewhat defined what that means (just read the law) and surprise: The way many datahogs wish it to be, isn't how the law was written.

E.g. if you trick or extort users into agreeing, consent was neither given informed nor freely. In front of the law it is as if you haven't asked for consent at all and GDPR fines can be up to 4% of the global turnover of the previous fiscal year. But yeah.


But then you don't have a company that can afford to build a 100b training cluster for example. If that becomes the new sota you are again left behind.


>If that becomes the new sota you are again left behind.

It's why Europe is being elft behind technologically compared to the US and also loosing worldwide share of GDP. Thousands of small local companies scattered across several countries can't compete with the scale the likes of FAANG can achieve in the US and worldwide.


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

Search: