The Irony of 'You Wouldn't Download a Car' Making a Comeback in AI Debates
The Irony of 'You Wouldn't Download a Car' Making a Comeback in AI Debates
Those claiming AI training on copyrighted works is "theft" misunderstand key aspects of copyright law and AI technology. Copyright protects specific expressions of ideas, not the ideas themselves. When AI systems ingest copyrighted works, they're extracting general patterns and concepts - the "Bob Dylan-ness" or "Hemingway-ness" - not copying specific text or images.
This process is akin to how humans learn by reading widely and absorbing styles and techniques, rather than memorizing and reproducing exact passages. The AI discards the original text, keeping only abstract representations in "vector space". When generating new content, the AI isn't recreating copyrighted works, but producing new expressions inspired by the concepts it's learned.
This is fundamentally different from copying a book or song. It's more like the long-standing artistic tradition of being influenced by others' work. The law has always recognized that ideas themselves can't be owned - only particular expressions of them.
Moreover, there's precedent for this kind of use being considered "transformative" and thus fair use. The Google Books project, which scanned millions of books to create a searchable index, was ruled legal despite protests from authors and publishers. AI training is arguably even more transformative.
While it's understandable that creators feel uneasy about this new technology, labeling it "theft" is both legally and technically inaccurate. We may need new ways to support and compensate creators in the AI age, but that doesn't make the current use of copyrighted works for AI training illegal or unethical.
For those interested, this argument is nicely laid out by Damien Riehl in FLOSS Weekly episode 744. https://twit.tv/shows/floss-weekly/episodes/744
You're viewing a single thread.
Are the models that OpenAI creates open source? I don't know enough about LLMs but if ChatGPT wants exemptions from the law, it result in a public good (emphasis on public).
Nothing about OpenAI is open-source. The name is a misdirection.
If you use my IP without my permission and profit it from it, then that is IP theft, whether or not you republish a plagiarized version.
So I guess every reaction and review on the internet that is ad supported or behind a payroll is theft too?
No, we have rules on fair use and derivative works. Sometimes they fall on one side, sometimes another.
Fair use by humans.
There is no fair use by computers, otherwise we couldn't have piracy laws.
The STT (speech to text) model that they created is open source (Whisper) as well as a few others:
Those aren't open source, neither by the OSI's Open Source Definition nor by the OSI's Open Source AI Definition.
The important part for the latter being a published listing of all the training data. (Trainers don't have to provide the data, but they must provide at least a way to recreate the model given the same inputs).
Data information: Sufficiently detailed information about the data used to train the system, so that a skilled person can recreate a substantially equivalent system using the same or similar data. Data information shall be made available with licenses that comply with the Open Source Definition.
They are model-available if anything.
I did a quick check on the license for Whisper:
Whisper's code and model weights are released under the MIT License. See LICENSE for further details.
So that definitely meets the Open Source Definition on your first link.
And it looks like it also meets the definition of open source as per your second link.
Additional WER/CER metrics corresponding to the other models and datasets can be found in Appendix D.1, D.2, and D.4 of the paper, as well as the BLEU (Bilingual Evaluation Understudy) scores for translation in Appendix D.3.
Whisper's code and model weights are released under the MIT License. See LICENSE for further details. So that definitely meets the Open Source Definition on your first link.
Model weights by themselves do not qualify as "open source", as the OSAID qualifies. Weights are not source.
Additional WER/CER metrics corresponding to the other models and datasets can be found in Appendix D.1, D.2, and D.4 of the paper, as well as the BLEU (Bilingual Evaluation Understudy) scores for translation in Appendix D.3.
This is not training data. These are testing metrics.
Edit: additionally, assuming you might have been talking about the link to the research paper. It's not published under an OSD license. If it were this would qualify the model.
I don't understand. What's missing from the code, model, and weights provided to make this "open source" by the definition of your first link? it seems to meet all of those requirements.
As for the OSAID, the exact training dataset is not required, per your quote, they just need to provide enough information that someone else could train the model using a "similar dataset".
Oh and for the OSAID part, the only issue stopping Whisper from being considered open source as per the OSAID is that the information on the training data is published through arxiv, so using the data as written could present licensing issues.
Ok, but the most important part of that research paper is published on the github repository, which explains how to provide audio data and text data to recreate any STT model in the same way that they have done.
See the "Approach" section of the github repository: https://github.com/openai/whisper?tab=readme-ov-file#approach
And the Traning Data section of their github: https://github.com/openai/whisper/blob/main/model-card.md#training-data
With this you don't really need to use the paper hosted on arxiv, you have enough information on how to train/modify the model.
There are guides on how to Finetune the model yourself: https://huggingface.co/blog/fine-tune-whisper
Which, from what I understand on the link to the OSAID, is exactly what they are asking for. The ability to retrain/finetune a model fits this definition very well:
The preferred form of making modifications to a machine-learning system is:
- Data information [...]
- Code [...]
- Weights [...]
All 3 of those have been provided.
From the approach section:
A Transformer sequence-to-sequence model is trained on various speech processing tasks, including multilingual speech recognition, speech translation, spoken language identification, and voice activity detection. These tasks are jointly represented as a sequence of tokens to be predicted by the decoder, allowing a single model to replace many stages of a traditional speech-processing pipeline. The multitask training format uses a set of special tokens that serve as task specifiers or classification targets.
This is not sufficient data information to recreate the model.
From the training data section:
The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages. As discussed in the accompanying paper, we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
This is also insufficient data information and links to the paper itself for that data information.
Additionally, model cards =/= data cards. It's an important distinction in AI training.
There are guides on how to Finetune the model yourself: https://huggingface.co/blog/fine-tune-whisper
Fine-tuning is not re-creating the model. This is an important distinction.
The OSAID has a pretty simple checklist for the OSAID definition: https://opensource.org/deepdive/drafts/the-open-source-ai-definition-checklist-draft-v-0-0-9
To go through the list of materials required to fit the OSAID:
Datasets Available under OSD-compliant license
Whisper does not provide the datasets.
Research paper Available under OSD-compliant license
The research paper is available, but does not fit an OSD-compliant license.
Technical report Available under OSD-compliant license
Whisper does not provide the technical report.
Data card Available under OSD-compliant license
Whisper provides the model card, but not the data card.
The problem with just shipping AI model weights is that they run up against the issue of point 2 of the OSD:
The program must include source code, and must allow distribution in source code as well as compiled form. Where some form of a product is not distributed with source code, there must be a well-publicized means of obtaining the source code for no more than a reasonable reproduction cost, preferably downloading via the Internet without charge. The source code must be the preferred form in which a programmer would modify the program. Deliberately obfuscated source code is not allowed. Intermediate forms such as the output of a preprocessor or translator are not allowed.
AI models can't be distributed purely as source because they are pre-trained. It's the same as distributing pre-compiled binaries.
It's the entire reason the OSAID exists:
- The OSD doesn't fit because it requires you distribute the source code in a non-preprocessed manner.
- AIs can't necessarily distribute the training data alongside the code that trains the model, so in order to help bridge the gap the OSI made the OSAID - as long as you fully document the way you trained the model so that somebody that has access to the training data you used can make a mostly similar set of weights, you fall within the OSAID
Edit: also the information about the training data has to be published in an OSD-equivalent license (such as creative Commons) so that using it doesn't cause licensing issues with research paper print companies (like arxiv)
OpenAI does not publish their models openly. Other companies like Microsoft and Meta do.