Science publisher Wiley is launching an artificial intelligence (AI) partnership program in an effort to “shape the AI world and the AI future, rather than be shaped by it.”
Josh Jarrett, senior vice president of AI growth at Wiley, announced the “co-innovation” partnership program at the Frankfurt Book Fair.
Wiley’s first partner is Potato, an AI research assistant “powered by peer-reviewed literature that enables high-quality science.” The partnership means Wiley and Potato are working together “to advance the application.”
Jarrett said The bookseller: “Potato is the first example in this broader partnership program, the Wiley AI partnership, co-innovation program that we will be announcing tomorrow. This is really part of our broader AI efforts.
“We really decided to lean on AI. We think this creates many opportunities and challenges, but that we can deal with them as effectively as possible by really being part of the solution, thinking along, experimenting, learning and sharing with our stakeholders. , to adjust. We really want Wiley and our partners to shape the AI world and the AI future, rather than be shaped by it.”
The partnership program will include the development of new AI applications, assistants and agents that leverage Wiley’s “rich authoritative content.”
In August, The bookseller reported that Wiley would earn a total of £33 million ($44 million) from two deals with unnamed tech companies in exchange for licensing the content for generative AI, without giving authors the opportunity to opt out. A Wiley spokesperson said at the time that “creating an opt-out or opt-in for a specific type of license on a case-by-case basis would falsely support AI developers’ misleading claim that licenses do not scale.”
Last week, Wiley said it had surveyed nearly 5,000 researchers and found that 69% of them wanted to keep up or be at the forefront of AI developments in their field. At the same time, the study highlighted that they need better and more reliable AI tools.
AI tools are known to sometimes “hallucinate” (make up facts that sound plausible but are imaginary), so making them more reliable is an important aspect of Wiley’s program.
Jarrett said The bookseller: “Accuracy, citation and transparency are critical principles that we strongly advocate in our work. The reality is that these fundamental large language models (LLM) are by definition not deterministic, right? They give you 10 suggestions. They come up with different ideas. Every time you ask them, they give you a slightly different answer. And if they can’t please you, they’ll hallucinate something.”
He added: “So for authoritative tasks we really need to refer to the source of the truth. So for applications like Potato, where we want to be able to perform really authoritative tasks, we want to use techniques more like RAG (retrieval augmented generation), using the power of the LLM that understands language and some basic reasoning. and creating generative content, but really, focus it only on this content, get your answers only from this authoritative content.”
He continued: “As publishers, we have all this rich, authoritative content. So when [our partners are] doing that synthesis of the research [via their tools]they want to be able to use the high-quality research published by Wiley to bring those answers back to researchers and avoid some of the challenges of hallucination.
Wiley is just one of many academic publishers that have entered into content licensing agreements or continue to license their authors’ backlists to train LLMs. Taylor & Francis and Wiley have already signed licensing deals, while Sage and University of Cambridge Press are considering this, with the latter seeking permission from authors first. Some academic authors have criticized publishers for entering into licensing agreements without first consulting authors or giving them the opportunity to opt out.
Jarrett says AI will revolutionize creativity and says it’s better to get involved than pretend it’s not happening.
He said: “People are already using it, so we can’t just close our eyes and pretend it’s not going to happen. We think a lot about AI-enhanced creativity. We will figure this out together. You know, in five years, most of the best research will probably have a combination of human-created thoughts and directions while using tools that are AI. We’re going to have to figure out how to combine these two things and operate within the right ethical frameworks and the right guidelines for universities, publishers and other stakeholders.”
