Hacker Newsnew | past | comments | ask | show | jobs | submit | motbus3's commentslogin

I will not go into the details of the topic but the "What to do" is the most obvious thing. If a paper that is impactful cannot be backed by other works that should be a smell

This is quite old news. I've heard about this more than 10 years ago at least. It has been fairly successful since the beginning and I've heard it improved quite a lot

Weird enough, they never asked Anthropic or Meta to do the same and restart the model training from scratch

Because Anthropic and Meta had lawyers, and Anna's Archive didn't even show up, so worldcat go whatever it demanded.

If it just forks chromium because it found it on the web it would also claim it made a browser from scratch. LLM does not know. It is not a person, it is a thing, just an algorithm

Wasn't there a meme called owl really?

I like the post but we can learn from insurance companies.

They have AI finding reasons to reject totally valid requests

They are putting to court that this is a software bug and they should not be liable.

That will be the standard excuse. I hope it does not work.


I know nothing about the topic. Although it seems a better alternative than coal or petrol, is it free of side effects for the nature? I wonder if the heat that would be spread around the atmosphere and back to space can actually gradually serve as a trap for heat?

Does this question make any sense at all?


No it doesn't make sense. Every photon that hits the Earth is eventually either absorbed as heat, reflected back into space or both (eg. partially absorbed and partially re-emitted as lower energy photons.) There is no net global increase in heat from a wind turbine or solar panel. (There might be slight local shifts.)

The only way this could change net heat if it significantly altered the reflectivity of the surface, and in practice the affected area is too small to matter. As an exaggerated example, I found an article [1] that calculated the area that would need to be covered by solar panels to generate power equal the total global electricity consumption to be 115,625 square miles, approximately equal to the state of New Mexico.

[1] https://www.axionpower.com/knowledge/power-world-with-solar/


This is actually quite a sizeable chunk. If in the future needs grow 10 times the area needed might become big problem.

It would actually be much better than nuclear. Remember, for every kWh of electrical energy delivered from a nuclear plant, 2 kWh of waste heat goes up those cooling towers. This is not the case with solar, particularly if it were built on ground that was already fairly dark.

Direct thermal pollution like this is not yet globally significant, but if demand increased to the point that land constraints actually applied then it would become important.


Might.

Sure, everything has downsides. Even breathing. But none of the alternatives have downsides that are as big as taking carbon from the soil and pumping it in an already stressed ecosystem.

> is it free of side effects for the nature?

What is free of side effects for "nature" ?


"You're absolutely right!" Those models are geared towards continuing the text. I have my impression that without that, the model would disagree much more as a chat/conversation

I feel the same. I understand some of the excitement. Whenn I use it I feel more productive as it seems I get more code done. But I never finish anything earlier because it never fails to introduce a bizarre bug or behaviour that no one in sane made doing the task would

They stole books, audio, image and tried to settle discussions with threats and closed party negotiations when they are caught.

What people expect from them?


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

Search: