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

I agree with this, and it really exposes how two faced and scandalous these companies are.

They all love green energy, but will they use that for their data centers in your backyard?

Rules for the and not for me !


I’m not against data centers, but they need to come with their own clean power supply, and export minimal pollution to its neighbors .

If you can’t provide that it’s not worth the downsides for the community.


Small power plants are less efficient and more polluting than larger ones. They are also harder to inspect and certify on a regular schedule. A single 1GW power facility in your county is much better than 10000 small power plants, one for each business.

True, centralized power generation using a grid that can carry that load would be the best solution. But in absence of that, the alternatives only seem to be between inefficient decentralized green and inefficient decentralized non-green solutions. In that case, I'd opt for the green ones. (Or of course: no datacenter at all)

Even for wind or solar ?

LOL. You’d need thousands of acres of solar panels to provide power for a single data center. Of course the exact area needed depends on what part of the world you’re in, but are you going to put up with 4 square miles of solar panels in your city? That’s a square two miles on a side, or 3.4km×3.4km. Or are you going to let them install hundreds of wind turbines in your neighborhood?

Yes. Don’t build the data center unless you can support that. What’s so strange about it?

Do you think ai usage will decrease due to this change ?


Yes. I've seen the AI usage patterns from people at the top of these kind of leaderboards, they understand that they are not doing things that they expect to produce commensurate value. At best they've inferred (sometimes correctly!) that the existence of the leaderboard means leadership has decided it's OK to burn prodigous amounts of token on any experimental thing one can imagine that might be useful.


> Do you think ai usage will decrease due to this change ?

Frivolous usage sure, internally at Amazon there is a subculture (if you can call it that) of award chasers. Using Kiro for mundane task to burn tokens does not sound that far-fetched.

Overall usage though no I don't think so, these tools have some pretty wide adoption at this point and not by people chasing awards.


Yeah, that is funny looking back. Books full of PII.

I always think when I send a work email I am also leaking my pii everywhere.


Eventually I would bet on ai using its own non human readable languages (brains?) to program in to reduce overhead.

It will be a black box, and the code will be generated just in time by ai for each api request


If the language is unreadable for humans, we really can't trust that it does what it claims to do, except by testing. This requires more trust in the system than is warranted IMO. You can never be sure that there's no "sleeper" logic waiting to get activated. See "Reflections on Trusting Trust" by Ken Thompson. If we build systems that start relying on opaque mechanisms, it seems to be only a matter of time before things start behaving in ways contrary to what their authors intended, with no clear way to stop it other than hitting the power button, if that's even possible at that point.


If that’s the case, are you saying the proof was part of the training set?

A lot of novelty is just gluing approaches together and reporting what sticks.


Yes.

A lot of brute force methods are the most inefficient means of solving a problem.

Language Models are a sign that our current human information infrastructure and access methods are completely wrong.


Arguably, since everything resolves to an axiom, isn’t the solution in any training set ?

The order and combination is what makes it special.

Is current human information access methods wrong, or do we just synthesize data in a way that is inefficient for this sort of problem solving ?


> since everything resolves to an axiom

This isn't true. There are solutions that are beyond apparent reason and logic. This is what a "breakthrough" is.

> The order and combination is what makes it special.

Given an infinite amount of time a team of monkeys will produce Shakespeare. Is that "special?" Perhaps we should leave some room for _how_ those combinations happen and how efficient they are.

> Is current human information access methods wrong

They are wrong. The largest search company is also the largest advertiser. I'm surprised that anyone either fails to apprehend this or pretends not to.


I’m a little confused. Pocket is outsourced to railway, which ended up deleting their data ?

I do find the author to be completely negligent , unless railway has completely lied about the safety in their product.


Are they practically doing this ?


When it comes to data quality it justifies a few checks.


I think someday it will be completely unreadable for humans. Ai will have its optimized form.


Cuba would be bombed off the map if that happened.


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

Search: