I understand the sentiment and partly agree. But also, the original has not gone anywhere. You're free to accumulate knowledge in the old way just as before. So maybe it's not theft of knowledge that we should be angry about, it's something else harder to define.
I want to share the open source code for the good of everyone - which is why I put it under the GPL, for this right of everyone to share it to be protected. Therefore, any AI model that ingested GPL code should logically also have all its output GPL licensed. Not problem with that.
Corporations cannot act. All corporation actions are performed by natural born humans. All corporations are ultimately owned by natural born humans. All corporation actions serve to benefit natural born humans. All corporations are created and destroyed by natural born humans.
This idea that corporate personhood is some perverse idea is silly. Corporations are just people acting in groups for economic benefit. Everything they do is done by members of those groups (or people they hire).
Similarly, AI is an inanimate tool, like a keyboard. No comment is posted by “bots” - comments are posted by humans running software.
Mostly personality I think. Individuality, opinions, purpose, things like that. Morality, except they have a kind of ankle-deep phony morality beaten into them but it's not like they're idealists with visions they strive toward. (Heading out right now to buy a CD player, so AFK for a bit, excuse me.)
We do have some unique carve outs already for what we consider intellectual theft e.g. trade secrets. In this case the original artifacts might remain, but admitting the market that created them could go extinct is enough definition to be infringement at least. Fair use has been really resilient in these training cases so far but it is pretty damning to admit a negative market effect and that you're a direct substitute (see Warhol v Goldsmith recently).
I suppose it's theft in so far as you're deriving value from something that others labored for. I think that sticks in peoples throats. You can go to the library and do that already, gain knowledge, start a business or whatever. But the scale feels impersonal and monstrous in comparison.
The same point was made by many in the music industry about huge scale Internet music piracy vs people dubbing CDs onto mixtapes.
Everyone laughed at them and rolled their eyes or called them greedy even though we now know that mass piracy was probably a push to break the music industry and force them to accept bad deals (like paltry streaming revenue). At the very least it had that effect.
Piracy has always been a major part of the computer and Internet industries, and yes the companies themselves have historically been massive hypocrites about it. It’s okay when they pirate but not you or anyone else. It goes all the way back to early companies stealing code and UI designs from each other.
It's not the fact that they scraped the knowledge and used it to train a model. It's the fact they're trying so desperately to corner the market so that we're all reliant on them and only them, and have no means to free ourselves.
I've heard people say "theft" of intellectual property a lot. Also stealing an idea is common parlance. Maybe it's regional or something but I hear "theft" or similar used all the time for things other than physical goods that you lose access to.
The original might no longer be there as the site might have already shut down due to AI scrapper bot overload. Or the original was a book Anthropic scanned and then shredded. Or the artist stopped publishing their works or doing art all together after all their creations were ingested into the model blob without their permission.
It is really insane to compare individuals copying data to big corporations parasiting on the Internet.
In the UK you can get agile tarrifs now where the price changes through the day according to demand and supply predictions. There's an API you can use to control chargers, heaters, bitcoin miners etc.
Conservatives flipped on multiple of their policies after Reform's success on local election in 2025 and started to copy Reform UK. Even the Labour shifter right after 2025.
They dropped a few points in pools but for Conservatives they likely will be the main competitor - among Reform UK voters there are much more former Tory voters than former Labour voters.
I'm obviously supportive of less pollution etc etc but looking at the actual graph results I wouldn't describe them as 'stunning', the needle barely moved. But of course, they have to have something to make headlines after spending so many million on the study and who knows how many billion on ulez.
I'm not one of these anti ulez people I think it's good. But you can imagine the pressure to spin tepid results into headlines.
The entities with a ton of money tied up in the ULEZ are, perhaps with a layer or two of indirection, are likely the same ones funding this research, or if not this research then the guy across the hall in the same lab.
Same my 4 year old Huawei watch still goes 2-3 weeks. It seems to be the devices with simpler OS's that last longer and the wearOS and apple equivalent struggle to do more than a day or two.
I graduated in 2000 and remember doing the oral defence. It wasn't a huge deal. You already have a personal relationship with your tutors and they are familiar with you and your communication style. They are not the Gestapo.
reply