This kind of thing is so avoidable by anyone who has not sold their soul. The answer is: if a company wants you to do a deal but requires as a condition that you not reveal to anyone that you are doing a deal with that company, you just say no. It's that simple.
FrontierMath was funded by OpenAI
201–210 of 212 posts
Re: FrontierMath was funded by OpenAI
#202Earlier quoted context omitted.
If the output of a mathematical model trained on an aggregate of knowledge that contains copyrighted material is derivative and infringing, then ipso facto, all works since the inception of copyright are derivative and infringing. You learned English, math, social studies, science, business, engineering, humanities, from a McGraw Hill textbook? Sorry, all creative works you’ve produced are derivative of your educatio…
> If the output of a mathematical model trained on an aggregate of knowledge that contains copyrighted material is derivative and infringing, then ipso facto, all works since the inception of copyright are derivative and infringing. I'm not saying every LLM output is necessarily infringing, I'm saying that some are, which means the underlying LLM (considered as a work on its own) must be. If you ask a human to come u…
You are saying that, any output is possibly infringing, dependandant on the input. This is actually, factually, verifiably, false, in terms of current copyright law.
No human, in the current epoch of education where copyright has been applicable, has learned, benefited, or exclusively created anything behreft of copyright. Please provide a proof otherwise if you truly believe so.
Re: FrontierMath was funded by OpenAI
#203Earlier quoted context omitted.
If the output of a mathematical model trained on an aggregate of knowledge that contains copyrighted material is derivative and infringing, then ipso facto, all works since the inception of copyright are derivative and infringing. You learned English, math, social studies, science, business, engineering, humanities, from a McGraw Hill textbook? Sorry, all creative works you’ve produced are derivative of your educatio…
I do appreciate your point because it's one of the interesting side effects of AI to me. Revealing just how much we humans are a stack of inductive reasoning and not-actually-free-willed rehash of all that came before. Of course, humans are also "trained" on their lived sensory experiences. Most people learn more about ballistics by playing catch than reading a textbook. When it comes to copyright I don't think the p…
My point isn’t adversarial, we most likely (in my most humble opinion) “learn” the same way as anything learns. That is to say, we are not unique in terms of understanding, “understandings”.
Are the ballistics we learn by physical interaction any different from the factual learning of ballistics that, for example, a squirrel learns, from their physical interactions?
Re: FrontierMath was funded by OpenAI
#204Earlier quoted context omitted.
> If the output of a mathematical model trained on an aggregate of knowledge that contains copyrighted material is derivative and infringing, then ipso facto, all works since the inception of copyright are derivative and infringing. I'm not saying every LLM output is necessarily infringing, I'm saying that some are, which means the underlying LLM (considered as a work on its own) must be. If you ask a human to come u…
You are saying that, any output is possibly infringing, dependandant on the input. This is actually, factually, verifiably, false, in terms of current copyright law. No human, in the current epoch of education where copyright has been applicable, has learned, benefited, or exclusively created anything behreft of copyright. Please provide a proof otherwise if you truly believe so.
> You are saying that, any output is possibly infringing, dependandant on the input.
What? No. How did you get that from what I wrote? Please engage with the argument I'm actually making, not some imaginary different argument that you're making up.
> No human, in the current epoch of education where copyright has been applicable, has learned, benefited, or exclusively created anything behreft of copyright.
What are you even trying to claim here?
Re: FrontierMath was funded by OpenAI
#205Earlier quoted context omitted.
> Copyright law is pretty clear on commissioned work, you are the holder, if your employee violated copyright and you failed to do your due diligence before publication, then you are responsible. No, for commissioned work in the usual sense the person you commissioned from is the copyright holder; you might have them transfer the copyright to you as part of your contract with them but it doesn't happen by default. It…
This is a common misconception. You mean the author. The creator of a commissioned work is the author under copyright law, the owner or copyright “holder” is the commissioner of the work or employer of the employee that created the work as a part of their job. The author may contractually retain copyright ownership per written agreement prior to creation, but this is not the default condition for commissioned, “speci…
> The creator of a commissioned work is the author under copyright law, the owner or copyright “holder” is the commissioner of the work or employer of the employee that created the work as a part of their job.
Nope. In cases where work for hire does apply (such as an employee preparing a work as part of their employment), the employer holds the copyright because they are considered as the author. But a work that's commissioned in the usual way (i.e. to a non-employee) is not a work-for-hire by default, in many cases cannot be a work-for-hire at all, and is certainly not a work-for-hire without written agreement that it is.
> The author may contractually retain copyright ownership per written agreement prior to creation, but this is not the default condition for commissioned, “specially ordered”, works
Nope. You must've misread this part of the law. A non-employee creator retains copyright ownership unless the work is commissioned and there is a written agreement that it is a work for hire before it is created (and it meets the categories for this to be possible at all).
> The only way an employer/commissioner would be responsible (vicarious liability) for copyright infringement of a commissioned work or work produced by an employee
What are you even trying to argue at this point? You've flipped to claiming the opposite of what you were claiming when I replied.
> duty of due diligence to ensure originality
This is just not a thing, not a legal concept that exists at all, and a moment's thought will show how impossible it would be to ever do. When someone infringes copyright, that person is liable for that copyright infringement. Not some other person who commissioned that first person to make something for them. That would be insane.
Re: FrontierMath was funded by OpenAI
#206Earlier quoted context omitted.
This is a common misconception. You mean the author. The creator of a commissioned work is the author under copyright law, the owner or copyright “holder” is the commissioner of the work or employer of the employee that created the work as a part of their job. The author may contractually retain copyright ownership per written agreement prior to creation, but this is not the default condition for commissioned, “speci…
> The creator of a commissioned work is the author under copyright law, the owner or copyright “holder” is the commissioner of the work or employer of the employee that created the work as a part of their job. Nope. In cases where work for hire does apply (such as an employee preparing a work as part of their employment), the employer holds the copyright because they are considered as the author. But a work that's co…
Quote the full passage of copyright law that backs any of your claims up.
Re: FrontierMath was funded by OpenAI
#207Earlier quoted context omitted.
> "I can't prove it, but I heard it from multiple people in the industry" The cited papers demonstrate that benchmark contamination exists as a general technical challenge, but are being misappropriated to support a much stronger claim about intentional misconduct by a specific actor. This is a textbook example of expanding evidence far, far, beyond its scope. > "The verbal agreement promised not to train on the eval…
> Attempting to justify potential misconduct through semantic technicalities ("well, validation isn't technically training") Validation is not training, period. I'll ask again: what is the possible goal of accessing the evaluation set if you don't plan to use it for anything except the final evaluation, which is what the test set is used for? Either they just asked for access without any intent to use the provided da…
> Validation is not training, period.
Sure, but what we care about isn't the semantics of the words, its the effects of what they're doing. Iterated validation plus humans doing hyperparameter tuning will go a long way towards making a model fit the data, even if you never technically run backprop with the validation set as input.
> OpenAI is not doing science; they are doing business.
Are you implying these are orthogonal? OpenAI is a business centered on an ML research lab, which does research, and which people in the research community have generally come to respect.
> at this point, the argument OpenAI did something rests on unfalsifiable claims about the industry as a whole, claiming insider knowledge, while avoiding any verifiable evidence.
No, it doesn't. What OP is doing is critiquing OpenAI for their misbehavior. This is one of the few levers we (who do not have ownership or a seat on their board) have to actually influence their future decisionmaking -- well-reasoned critiques can convince people here (including some people who decide whether their company uses ChatGPT vs. Gemini vs. Claude vs. ...) that ChatGPT is not as good as benchmarks might claim, which in effect makes it more expensive for OpenAI to condone this kind of misbehavior going forward.
The argument that "no companies are moral, so critiquing them is pointless" is just an indirect way of running cover for those same immoral companies.
Re: FrontierMath was funded by OpenAI
#208Earlier quoted context omitted.
> The creator of a commissioned work is the author under copyright law, the owner or copyright “holder” is the commissioner of the work or employer of the employee that created the work as a part of their job. Nope. In cases where work for hire does apply (such as an employee preparing a work as part of their employment), the employer holds the copyright because they are considered as the author. But a work that's co…
Quote the full passage of copyright law that backs any of your claims up.
"(2) a work specially ordered or commissioned for use as a contribution to a collective work, as a part of a motion picture or other audiovisual work, as a translation, as a supplementary work, as a compilation, as an instructional text, as a test, as answer material for a test, or as an atlas, if the parties expressly agree in a written instrument signed by them that the work shall be considered a work made for hire. For the purpose of the foregoing sentence, a “supplementary work” is a work prepared for publication as a secondary adjunct to a work by another author for the purpose of introducing, concluding, illustrating, explaining, revising, commenting upon, or assisting in the use of the other work, such as forewords, afterwords, pictorial illustrations, maps, charts, tables, editorial notes, musical arrangements, answer material for tests, bibliographies, appendixes, and indexes, and an “instructional text” is a literary, pictorial, or graphic work prepared for publication and with the purpose of use in systematic instructional activities.
In determining whether any work is eligible to be considered a work made for hire under paragraph (2), neither the amendment contained in section 1011(d) of the Intellectual Property and Communications Omnibus Reform Act of 1999, as enacted by section 1000(a)(9) of Public Law 106–113, nor the deletion of the words added by that amendment—
(A) shall be considered or otherwise given any legal significance, or
(B) shall be interpreted to indicate congressional approval or disapproval of, or acquiescence in, any judicial determination,
by the courts or the Copyright Office. Paragraph (2) shall be interpreted as if both section 2(a)(1) of the Work Made For Hire and Copyright Corrections Act of 2000 and section 1011(d) of the Intellectual Property and Communications Omnibus Reform Act of 1999, as enacted by section 1000(a)(9) of Public Law 106–113, were never enacted, and without regard to any inaction or awareness by the Congress at any time of any judicial determinations."
Now your turn, quote the full passage of whatever law you think creates this "duty of due diligence" that you've been talking about.
Re: FrontierMath was funded by OpenAI
#209Tim Gowers, one of the Fields medallists contributed problems to the benchmark dataset isn't happy about being misled about OpenAIs involvement. He retweeted this: https://x.com/Mihonarium/status/1880944026603376865?t=QN3i_X...
Re: FrontierMath was funded by OpenAI
#210Earlier quoted context omitted.
Quote the full passage of copyright law that backs any of your claims up.
"(2) a work specially ordered or commissioned for use as a contribution to a collective work, as a part of a motion picture or other audiovisual work, as a translation, as a supplementary work, as a compilation, as an instructional text, as a test, as answer material for a test, or as an atlas, if the parties expressly agree in a written instrument signed by them that the work shall be considered a work made for hire…
> b) Works Made for Hire.
>In the case of a work made for hire, the employer or other person for whom the work was prepared is considered the author for purposes of this title, and, unless the parties have expressly agreed otherwise in a written instrument signed by them, owns all of the rights comprised in the copyright.
https://www.copyright.gov/title17/92chap2.html#201
You are responsible for infringing works you publish, whether they are produced by commission or employee.
Due diligence refers to the reasonable care, investigation, or steps that a person or entity is expected to take before entering into a contract, transaction, or situation that carries potential risks or liabilities.
Vicarious copyright infringement is based on respondeat superior, a common law principle that holds employers legally responsible for the acts of an employee, if such acts are within the scope and nature of the employment.