I believe that new methods of distributing wealth will be necessary in the face of generative AI's effect. This proposal makes a good deal of sense to me. But is it really public data? Isn't most of the data on the internet owned (at least according to ToS) by the company that runs the platform that serves it up? A comment on Reddit is (non-exclusively) owned by Advance Publications, for instance. We all willingly put our data on websites and social media to be consumed by human and computer alike.
It's a good question. Right now AI companies are using the data under "fair use" and they may get away with it so Reddit or whomever can't claim it either. (Getty Images and the like are going after generative AI not for using training data, but for output.)
This proposed policy would say "well, if you're going to use public data for this use, you have to pay a fee because it's really all our data". Basically it's a mandatory licensing fee.
It's certainly going to be interesting to see how it all pans out! I suppose a useful tweak to the proposal might be that the licensing fee is only owned if the result isn't open source.
Well, it was public for anyone to read - that's inherently what you have done when you put something on the public web.
As far as GPT using it, that could arguably be considered "transformative use", and therefore be allowed under copyright law. (Note well: IANAL. And as far as I know, no court has ruled on this question so far, so your actual mileage may vary.)
So it may well work out that it's "public enough" for this use.
As for distributing wealth, I could argue that we (re)distribute wealth far more after the industrial revolution than we did before it. A huge increase in production enabled (and, in at least some eyes, justified) an increase in distribution. AI may do the same.
But personally, I don't buy the "we're all going to be unemployed" theory. The industrial revolution didn't make people unemployed; they just worked in a factory instead of on a farm. COBOL didn't make professional programmers unemployed (despite that being the design goal!); it made them COBOL programmers instead of assembly programmers. In the same way, AI is a lever that will make those that use it more productive, but the result will be that we produce more, rather than that fewer of us produce at all.
After years of worrying about the worst-case scenarios for climate change and then Covid, I ran out of worry. I'm nothing but excited about generative AI, even if it does take my job. Beyond the financial anxiety, I wouldn't want to do a job that AI could do as well because of a law. That's just mandated busy work.
10 comments
[ 2.6 ms ] story [ 33.8 ms ] threadThis proposed policy would say "well, if you're going to use public data for this use, you have to pay a fee because it's really all our data". Basically it's a mandatory licensing fee.
As far as GPT using it, that could arguably be considered "transformative use", and therefore be allowed under copyright law. (Note well: IANAL. And as far as I know, no court has ruled on this question so far, so your actual mileage may vary.)
So it may well work out that it's "public enough" for this use.
As for distributing wealth, I could argue that we (re)distribute wealth far more after the industrial revolution than we did before it. A huge increase in production enabled (and, in at least some eyes, justified) an increase in distribution. AI may do the same.
But personally, I don't buy the "we're all going to be unemployed" theory. The industrial revolution didn't make people unemployed; they just worked in a factory instead of on a farm. COBOL didn't make professional programmers unemployed (despite that being the design goal!); it made them COBOL programmers instead of assembly programmers. In the same way, AI is a lever that will make those that use it more productive, but the result will be that we produce more, rather than that fewer of us produce at all.
One problem is everything is copyrighted.
$0.001 per word