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

Maybe they can use the abundant oil for cooling as well :) that is how transformers are cooled

I was thinking heat pumps hooked up to radiators and/or geo grids. Right now the cheapest method is evaporators which use a ton of water. It's expensive to get rid of a couple megawatts like that. Some data centers have their own microclimates that are 10-20C hotter on average.

funny you say this, my meta swe friends have been talking about how many more loc they have been producing since they were given llms tens of thousands more. the thing i have found most interesting is how that seems to be the only metric that actually matters to them. just generating more and more.

Give a man a hammer...and tell him his performance review outcome is tied to his use of that hammer -> then watch every problem being, for the lack of a better term, nailed down . With the said hammer of course.

Facebook and Oracle also build their own, at least before the last couple years where they’ve financed out to new bag holders.

I work for a completely unrelated fed agency, who doesn’t use Anthropic products, and we all received the email stating we couldn’t use them period.

Huh, does supply-chain risk mean SecDef can bar a company from all federal contracting?

I have no idea but this went out to all fed agencies from what I could tell looking at the subreddit for fed employees. I was surprised by the notice because my agency does not have a contract with them and obviously we can’t just use any LLM provider.

Correct. And this quickly expands out into most companies in the US as the federal government uses and buys a huge amount of software. A component that you make and sell to X, that is used in Y, which is bundled up in Z that had Anthropic used on it can't be used by the fed.gov.

Due-diligence once again proving to be mostly cargo-culting.


I’m still learning this advantages and differences between them, would there be benefits to SFT and RAG? Or does RAG make SFT redundant?


I think generally, SFT is like giving the LLM increased intuition in specific areas. If you combine this with RAG, it should improve the performance or accuracy. Sort of like being a lawyer and knowing something is against the law by intuition, but needing the library to cite a specific case or statute as to why.


Thank you I appreciate the reply and that analogy helps make sense of this.


Why wouldn’t an advanced AGI robot, trained on human behavior, not want their own house and mode of transportation? Sure it’s basically kayfabe for them to ‘want’ the stuff we do but if we’re following the script of who will buy all the stuff, then the answer will be the robots I guess.

You think housing market are tough now, wait until you’re competing with 5 robot families who all have jobs you used to do.


You can break it down by, new construction, planned renovations/improvements, and emergency repairs.

Not everyone works all three or wants to do more than one of these groups. There’s different levels of demand, pay, competition at each.


Instead of slapping on an extra battery pack, it will be an onboard llm model. Could have lifecycles just like phones.

Getting bigger (foldable) phones, without losing battery life, and running useable models in the same form-factor is a pretty big ask.


I could see a future integrated feature among cars brands doing this with their cameras. Teslas in your area on the road are already monitoring everything, then flag available street parking spots so autopilot cars looking for parking are dispatched to the nearest one. No need for drones if you already have a sufficiently large fleet of cars on the road.


Why wouldn't we just wire up all the parking spaces with sensors and then have a city wide parking app just for this? That seems far more efficient and safer over having a fleet of parking scout drones randomly flying throughout the city.


Because the car owner has an incentive to spend a couple thousand extra for a scout drone, but cities have very little incentive to spend many millions on a major infrastructure project that achieves the same. Despite that being cheaper than everyone getting a drone port in their car


It feels intuitive that cities would have the same incentive to maximize use of parking spots as they do having them in the first place. Cities want commerce to flow.

But let’s simplify it down a bit further: pretend all parking spots are for-profit. These lots would want to communicate vacancy to maximize use. Much like how motels are motivated to tell you when they have vacancy without you having to stop to find out.


The correct answer seems to be: because our economic model requires that new junk be constantly invented in order that we have something to buy. Alas.


Consider applying for YC's Summer 2026 batch! Applications are open till May 4

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

Search: