[License-review] License-review Digest, Vol 117, Issue 2
Emanuele D.
deboni.emanuele at gmail.com
Tue Apr 18 14:17:51 UTC 2023
Hello there,
First of all thanks for replying, Your replies are giving me a lot of useful knowledge
and insights on how to eventually work on this failing attempt. I will answer as following:
———
Dear @mirabilos,
> I call absolute, utter BULLSHIT on this.
The idea, whether possible, as explained, is to restrict AI entities (operating autonomous agent) not the AI source code nor the AI System. Generally speaking:
1) a open source developer would be able to build (according to the relative OS license) any AI source code, model or system. These are tools of the open source community, and they fall under the OSD.
2) an AI Entity, born by that tool (without entering in ethics), and generally speaking more or less autonomous, it’s not a human developer nor a tool. Therefore there should be more attention from a legal point of view.
The non-deterministic nature is given by the probabilistic pattern that powers machine learning, as you might know, unrelated to the deterministic nature of the computers, this is a fact.
It’s like saying that since we know your DNA, we can tell what you are going to do in a sunny day of June (of course this is a exaggeration, but I hope you get the point).
Talking, for example, about GPT models, if you ask “How much is the result of 5 * 3?” for three times, it will give you three different answers more or less about the fact that the result it 15.
There are already many resources on it, I will give you one among many, that you might already know:
- https://www.linkedin.com/pulse/v-word-challenge-ai-determinism-part-2-charles-chen/ <https://www.linkedin.com/pulse/v-word-challenge-ai-determinism-part-2-charles-chen/>
At the moment, the 5 * 3 question might look irrelevant as covered case, it looks like conventional software. But let’s take a look at the same exact GPT model, applied in a different way, in this experiment (already usable now):
- https://github.com/Significant-Gravitas/Auto-GPT <https://github.com/Significant-Gravitas/Auto-GPT>
The problem here is that if you ask the AI to go from X to Z, you don’t know what the exact steps will be in the middle [X —> unknown —> unknown(n) —> Z(ish)], and that should raise a flag for the community.
Because, even if you can reliably bet on a Z(ish) result, you can’t predict it with 100% of probability, and you can’t tell what will be done precisely in order to get there.
In my opinion. this brings a problem of accountability and the OSD lacks to identify this distinction.
Who will be accountable for any legal issue? Also here many resources start to popup, for example https://arxiv.org/pdf/2206.11834.pdf <https://arxiv.org/pdf/2206.11834.pdf>.
———
Dear Carlo Pina,
> I must confess I have many troubles with your "license", starting from the fact that it is not a license, but a restriction applied on the top of a license. As the name suggests, the main intent of this document is to impose a restriction on certain uses, users and fields of endeavour: for that it is directly against #5 and #6, in my reading. That alone disqualifies the "license". Have you spent time considering the OSD before applying for approval, in the face of this striking problem, without even trying to address it with a convincing rationale?
Yes, like I mentioned in the text, point 5 and 6 are referring to “persons” not AI autonomous agents, which shouldn’t be falling in neither definition of software and persons.
However, you are very right, this is my limitation and lack of legal knowledge about the difference between license and restriction. I will definitely ask for a legal advice on this.
> Frankly, and this is perhaps the most fatal flow, I am not buying the very concept of "let's keep open source from AI generators" at all.
Please, consider the fact, as before mentioned, that these are not just “generators”.
> Additionally, you claim that the SPDX (not SPIDX) identifier for your "license" is RAIL, but I see no trace of it in the official list of licenses. In order to claim it's an SPDX id, you should ask for it in the first place and since it is basically a first-come-first-served basis, using it before having secured it is not a good idea.
Oh! I didn’t know about this. I thought this was assigned after an eventual approval, and that should be ask in the submission before.
> I urge you to reconsider your submission -- which I think should be outright rejected without much further ado -- and the very underlying purpose. And to consider it in the light of the legislative initiatives which -- unlike open source licensing -- have some chances to have an impact on the good handling of AI.
Thanks for the advice. I will try to do my part.
———
Dear Simon,
Thanks for the advice as well.
———
Best regards.
> On 18 Apr 2023, at 14:00, license-review-request at lists.opensource.org wrote:
>
> Send License-review mailing list submissions to
> license-review at lists.opensource.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
>
> or, via email, send a message with subject or body 'help' to
> license-review-request at lists.opensource.org
>
> You can reach the person managing the list at
> license-review-owner at lists.opensource.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of License-review digest..."
>
>
> Today's Topics:
>
> 1. Re: Approval - Restricted Artificial Intelligence License
> (RAIL) (Thorsten Glaser)
> 2. Re: Approval - Restricted Artificial Intelligence License
> (RAIL) (Carlo Piana)
> 3. Re: Approval - Restricted Artificial Intelligence License
> (RAIL) (Simon Phipps)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Tue, 18 Apr 2023 09:54:56 +0000 (UTC)
> From: Thorsten Glaser <tg at mirbsd.de>
> To: License submissions for OSI review
> <license-review at lists.opensource.org>
> Subject: Re: [License-review] Approval - Restricted Artificial
> Intelligence License (RAIL)
> Message-ID: <Pine.BSM.4.64L.2304180945300.10778 at herc.mirbsd.org>
> Content-Type: TEXT/PLAIN; charset=utf-8
>
> Emanuele D. dixit:
>
>> AI entities can learn, adapt, and perform complex tasks *autonomously*
>> and with *non-deterministic* patterns.
>
> I call absolute, utter BULLSHIT on this.
>
> AI is software that runs on deterministic computers, PRNG input
> notwithstanding. It?s equivalent to a compiler, and its output
> is not ?generated? (it wouldn?t need any ?training data? if it
> were so) but a derivative of its input (training data, and to a
> small amount prompt and possibly PRNG), in a deterministic way.
>
> So-called ?AI? cannot be the subject of copyright because it?s
> not a juristic entity. (AFAIR this was already confirmed in
> court: there can be no copyright claims on AI output.)
> Therefore, any copyright-relevant issues on AI is bound to its
> users as it?s just a tool. I maintain that in almost all cases,
> and generally enough to assume this true unless proven otherwise
> (via an auditable code and ?training data? trail) its output is
> n?n-distributable because that violates the licence on the input.
> (It has been proven that the input can even be recovered in large
> enough amounts to be recognisable, about ? of an image complete
> with watermarks for example.)
>
> I believe we don?t need any specific licences for this, and that
> the existing ones are not only sufficient (they specify conditions
> on creating derivative works and ?conveying? them) but also the
> maximum protection possible (anything further would be a restriction
> on use (by the legal entity operating the AI software / ?compiler?)
> which is not free). I did add an interpretation on the MirOS Licence
> webpage explaining this; use of ?AI? where the input is homogenously
> licenced and all copyright and licence statements are retained as
> required, if it existed, would be legal.
>
> bye,
> //mirabilos
> --
> I believe no one can invent an algorithm. One just happens to hit upon it
> when God enlightens him. Or only God invents algorithms, we merely copy them.
> If you don't believe in God, just consider God as Nature if you won't deny
> existence. -- Coywolf Qi Hunt
>
>
>
> ------------------------------
>
> Message: 2
> Date: Tue, 18 Apr 2023 12:55:44 +0200 (CEST)
> From: Carlo Piana <osi-review at piana.eu>
> To: License submissions for OSI review
> <license-review at lists.opensource.org>
> Subject: Re: [License-review] Approval - Restricted Artificial
> Intelligence License (RAIL)
> Message-ID:
> <507567272.10021936.1681815344637.JavaMail.zimbra at piana.eu>
> Content-Type: text/plain; charset=utf-8
>
> Dear Emanuele,
>
> I must confess I have many troubles with your "license", starting from the fact that it is not a license, but a restriction applied on the top of a license. As the name suggests, the main intent of this document is to impose a restriction on certain uses, users and fields of endeavour: for that it is directly against #5 and #6, in my reading. That alone disqualifies the "license". Have you spent time considering the OSD before applying for approval, in the face of this striking problem, without even trying to address it with a convincing rationale?
>
> I highlight the relevant language in ?3:
>
>> The Restricted Artificial Intelligence License, or simply RAIL, **is a license extension **
>> designed to **restrict** the [permission to] use, copy, modify, merge, publish, distribute, sublicense of
>> software under an existing open source license, for AI entities.
>
> There are many other shortcomings that appear abundantly evident. The language is quirky, it is very difficult to understand what is the legal implication of applying it to any license (eg. what happens to the GPLv3 under section 7?), I reckon it has not been vetted by a lawyer. The fact that you call it a license, when a license is in its essence a permission and you are proposing something that adds a restriction to a permission, does not give reassurances in the fact that you understood what is the subject matter of "license approval":
>
> Additionally, you claim that the SPDX (not SPIDX) identifier for your "license" is RAIL, but I see no trace of it in the official list of licenses. In order to claim it's an SPDX id, you should ask for it in the first place and since it is basically a first-come-first-served basis, using it before having secured it is not a good idea.
>
> Frankly, and this is perhaps the most fatal flow, I am not buying the very concept of "let's keep open source from AI generators" at all.
>
> I urge you to reconsider your submission -- which I think should be outright rejected without much further ado -- and the very underlying purpose. And to consider it in the light of the legislative initiatives which -- unlike open source licensing -- have some chances to have an impact on the good handling of AI.
>
> I'm replying as a FOSS licensing expert, not as member of the Board of OSI. This reply is therefore not part of a formal process within OSI.
>
> With best regards,
>
> Carlo Piana
>
>
> ----- Messaggio originale -----
>> Da: "Emanuele D." <deboni.emanuele at gmail.com>
>> A: "license-review at lists.opensource.org" <license-review at lists.opensource.org>
>> Inviato: Luned?, 17 aprile 2023 17:33:12
>> Oggetto: [License-review] Approval - Restricted Artificial Intelligence License (RAIL)
>
>> Dear Board,
>
>> I hope you are well and safe. My name Is Emanuele De Boni, I am a philologist by
>> academic education turned into software engineer. My Master?s thesis was around
>> open source licenses (2012) and since then I?ve been part (passively) of the
>> community.
>
>> I recently reflected about the importance of a clear distinction between humans
>> and AI entities, to push the Open Source Initiative forward and protect the
>> drive to innovation, collaboration and creativity that resulted from its
>> principles.
>
>> As a specialised linguistic, I would like here to submit at your attention the
>> attached license draft, containing already the supporting data reminded here [
>> https://opensource.org/licenses/review-process/#approval |
>> https://opensource.org/licenses/review-process/#approval ] . I give also, my
>> availability to call and discuss it together, I am currently located in
>> Amsterdam Time Zone (CEST).
>
>> I attached 3 textual file types: .txt, .md, .pdf.
>
>> Looking forward,
>> Kind Regards.
>
>> Emanuele De Boni
>> +31657175374 (active on WhatsApp and Telegram)
>> [ https://www.linkedin.com/in/emanueledeboni/ |
>> https://www.linkedin.com/in/emanueledeboni/ ]
>
>> _______________________________________________
>> The opinions expressed in this email are those of the sender and not necessarily
>> those of the Open Source Initiative. Communication from the Open Source
>> Initiative will be sent from an opensource.org email address.
>
>> License-review mailing list
>> License-review at lists.opensource.org
>> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
>
>
>
> ------------------------------
>
> Message: 3
> Date: Tue, 18 Apr 2023 12:21:52 +0100
> From: Simon Phipps <simon at webmink.com>
> To: License submissions for OSI review
> <license-review at lists.opensource.org>
> Subject: Re: [License-review] Approval - Restricted Artificial
> Intelligence License (RAIL)
> Message-ID:
> <CAA4ffp_Ejqh6tuxVo2ZWrk5e+icVufL6pDgL-iqOhzhawXHusA at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> Hi all,
>
> I do not believe this submission represents a valid license, since it is a
> manifesto for an unspecified text intended to be applied to modify
> unspecified open source licenses. Since we do not have the full text of any
> combination to consider for approval, I believe this submission should be
> automatically rejected.
>
> Best regards,
>
> Simon
> (in a personal capacity)
>
> On Tue, Apr 18, 2023 at 10:32?AM Emanuele D. <deboni.emanuele at gmail.com>
> wrote:
>
>> Dear Board,
>>
>> I hope you are well and safe. My name Is Emanuele De Boni, I am a
>> philologist by academic education turned into software engineer. My
>> Master?s thesis was around open source licenses (2012) and since then I?ve
>> been part (passively) of the community.
>>
>> I recently reflected about the importance of a clear distinction between
>> humans and AI entities, to push the Open Source Initiative forward and
>> protect the drive to innovation, collaboration and creativity that resulted
>> from its principles.
>>
>> As a specialised linguistic, I would like here to submit at your attention
>> the attached license draft, containing already the supporting data reminded
>> here https://opensource.org/licenses/review-process/#approval. I give
>> also, my availability to call and discuss it together, I am currently
>> located in Amsterdam Time Zone (CEST).
>>
>> I attached 3 textual file types: .txt, .md, .pdf.
>>
>>
>>
>> Looking forward,
>> Kind Regards.
>>
>> Emanuele De Boni
>> +31657175374 (active on WhatsApp and Telegram)
>> https://www.linkedin.com/in/emanueledeboni/
>>
>> _______________________________________________
>> The opinions expressed in this email are those of the sender and not
>> necessarily those of the Open Source Initiative. Communication from the
>> Open Source Initiative will be sent from an opensource.org email address.
>>
>> License-review mailing list
>> License-review at lists.opensource.org
>>
>> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
>>
>
>
> --
> *Simon Phipps*
> *Office:* +1 (415) 683-7660 *or* +44 (238) 098 7027
> *Signal/Telegram/Mobile*: +44 774 776 2816
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <http://lists.opensource.org/pipermail/license-review_lists.opensource.org/attachments/20230418/96476b1c/attachment-0001.html>
>
> ------------------------------
>
> Subject: Digest Footer
>
> _______________________________________________
> License-review mailing list
> License-review at lists.opensource.org
> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
>
>
> ------------------------------
>
> End of License-review Digest, Vol 117, Issue 2
> **********************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensource.org/pipermail/license-review_lists.opensource.org/attachments/20230418/0c511471/attachment-0001.html>
More information about the License-review
mailing list