16 Artificial Intelligence Part 4: AI Negotiation Case Study

In Part 4, we thought it would be helpful to share an example of AI restrictions that publishers proposed, followed by the outcomes for which we negotiated.

Publisher Proposal

The publisher’s first paragraph forbids the training or improving of any AI tool if it’s accessible or released to third parties. And it forbids the use of any outputs or analysis derived from the content to train any tool available to third parties. What does this mean? It would mean that the trained tools we explored in Parts 1-3 of our AI discussions would never be able to be released to third parties.

The paragraph is perhaps even more concerning. It provides that when using third party AI tools of any kind, scholars can use only limited portions of the licensed content with them, and can’t do any training at all of third party tools even if it’s a non-generative tool and they’re doing it in a completely closed environment! In other words, this proposal wouldn’t let scholars train a third party tool under any circumstances, much less make that tool available to others. It would in essence constitute a flat ban on nearly every research project we’ve explored in these chapters.

Response to Publisher

We explained to the publisher that within the University of California, we have a strong mandate from the president to obtain rights for our scholars to better the world using modern research technologies. In addition, the European Union’s Copyright Digital Single Market Directive and recent AI Act nullify any attempt to circumscribe the text and data mining and AI training rights reserved for scientific research within research organizations. So if the publishers have to agree to these terms in Europe, they can do so here.

Finally, we explained the many kinds of research projects that would be precluded by the publisher’s proposal. For instance:

Tools that could not be disseminated

1. In 2017, chemists created and trained a generative AI tool on 12,000 published research papers regarding synthesis conditions for metal oxides, so that the tool could identify anticipated chemical outputs and reactions for any given set of synthesis conditions entered into the tool. The generative tool they created is not capable of reproducing or redistributing any licensed content from the papers; it has merely learned conditions and outcomes and can predict chemical reactions based on those conditions and outcomes. And this beneficial tool would be prohibited from dissemination under the publisher’s terms identified above.
2. In 2018, researchers trained an AI tool (that they had originally created in 2014) to understand whether a character is “masculine” or “feminine” by looking at the tacit assumptions expressed in words associated with that character. That tool can then look at other texts and identify masculine or feminine characters based on what it knows from having been trained before. The implications are that scholars can therefore use texts from different time periods with the tool to study representations of masculinity and femininity over time. No licensed content, no licensed or copyrighted books from a publisher can ever be released to the world by sharing the trained tool; the trained tool is merely capable of topic modeling—but the publisher’s above language would prohibit its dissemination nevertheless.

Tools that could neither be trained nor disseminated 

1. In 2019, authors used text from millions of books published over 100 years to analyze cultural meaning. They did this by training third-party non-generative AI word-embedding models called Word2Vec and GLoVE on multiple textual archives. The tools cannot reproduce content: when shown new text, they merely represent words as numbers, or vectors, to evaluate or predict how similar words in a given space are semantically or linguistically. The similarity of words can reveal cultural shifts in understanding of socioeconomic factors like class over time. But the publisher’s above licensing terms would prohibit the training of the tools to begin with, much less the sharing of them to support further or different inquiry.
2. In 2023, scholars trained a third-party-created open-source natural language processing (NLP) tool called Chemical Data Extractor (CDE). Among other things, CDE can be used to extract chemical information and properties identified in scholarly papers. In this case, the scholars wanted to teach CDE to parse a specific type of chemical information: metal-organic frameworks, or MoFs. Generally speaking, the CDE tool works by breaking sentences into “tokens” like parts of speech and referenced chemicals. By correlating tokens, one can determine that a particular chemical compound has certain synthetic properties, topologies, reactions with solvents, etc. The scholars trained CDE specifically to parse MoF names, synthesis methods, inorganic precursors, and more—and then exported the results into an open source database that identifies the MoF properties for each compound. Anyone can now use both the trained CDE tool and the database of MoF properties to ask different chemical property questions or identify additional MoF production pathways—thereby improving materials science for all. Neither the CDE tool nor the MoF database reproduces or contains the underlying scholarly papers that the tool learned from. Yet, neither the training of this third-party CDE tool nor its dissemination would be permitted under the publisher’s restrictive licensing language cited above.

Accordingly, we marked up the publisher’s language to the extent it fails to confer essential text and data mining and AI usage, training, and dissemination rights. We also conveyed that our changes should appear familiar as the publisher is obligated to honor these same rights with European research institutions under Article 3 of the Copyright Digital Single Market Directive, and the recent AI Act.

We then summarized for the publisher the effect of our changes: