Not going to lie, you’re not missing out on much. The Atlantic for some reason has pushed 3-4 of these stories recently but all of them are very opinionated and not focusing on any specific cause, just kind of generalization that isn’t all that useful.
AFAIK, archive.ph does funky stuff with DNS to give them time to respond to take down requests [0]. If you've made any changes to what DNS you use, you might want to either revert them or setup split DNS to pass requests for archive.ph along accordingly
It is about time a creator respects The Commons in the AI era. Their mind did not grow in a vacuum -- they read billboards, handbills, novels from the public library, books passed to them in high school English class and so on. All of us build sentences out of words that precede us and have evolved and been shared for millennia. This interplay and exchange continues in the AI age. To shout "theft" is to dream and cement the path toward a "pay per thought" society. If there is ever to be venal economic exchange for AI produce, levy it upon model outputs, never on its fair use inputs. Otherwise, be honest and demand that we all ought to pay whenever we read a stop sign, or pay whenever we are forced to hear a song we absolutely hate when walking along a storefront.
Well, yes, and we can take this further to paying no one especially for his being in the right place and the right time to take credit for his community's investment in him.
Models are _not_ human beings, and should _not_ be treated as such! Not in ethical, normative, or legal fashions.
> If there is ever to be venal economic exchange for AI produce, levy it upon model outputs
Every output should be considered a derivative work of the input, and models should be required to license their inputs accordingly. To do otherwise is a massive devaluation of real human labour and ingenuity.
And no, this is not equivalent to human beings training and learning on their own experiences and consumption. Moreover, the scale of investing inputs and delivering outputs far outstrips any human capacity.
If human learning is equivalent to ML training; it’s up to the ML advocates to actually prove that! Until then, kindly stop using human comparisons as an excuse for rampant theft.
10 comments
[ 0.22 ms ] story [ 276 ms ] thread[0] https://news.ycombinator.com/item?id=36970702
Models are _not_ human beings, and should _not_ be treated as such! Not in ethical, normative, or legal fashions.
> If there is ever to be venal economic exchange for AI produce, levy it upon model outputs
Every output should be considered a derivative work of the input, and models should be required to license their inputs accordingly. To do otherwise is a massive devaluation of real human labour and ingenuity.
And no, this is not equivalent to human beings training and learning on their own experiences and consumption. Moreover, the scale of investing inputs and delivering outputs far outstrips any human capacity.
If human learning is equivalent to ML training; it’s up to the ML advocates to actually prove that! Until then, kindly stop using human comparisons as an excuse for rampant theft.
Your entire post is disingenuous and full of assumptions, like AI inputs are inherently fair use.