Historically, AI applications at Silverchair were limited to specific projects or features, but as the scope of possibilities expanded, we recognized the need to invest in AI experimentation and prototyping. A group of technologists, including me and other developers, started meeting regularly to build up our own knowledge of the inner workings of the technology and start building AI-driven applications to see what the possibilities were. Over time, our experiments gained traction, leading to the formation of the AI Team a few months later. The team has grown since then into a five-person group, which is not only focused on developing prototypes but also on providing industry guidance and exploring the path to productization.
Our team’s efforts have taken us to numerous conferences, beginning with NISO Plus and extending to STM, CESSE, and beyond. We’ve also hosted webinars and working groups to discuss the implications of AI in publishing, exploring both its potential benefits and risks. These events have offered invaluable insights into how AI is being received by the industry and have highlighted the importance of balancing innovation with practical, value-driven applications. Before diving into what I’ve observed and learned at these events, I want to review the technological changes that I have seen take place.
The Changing Landscape of AI Applications
When I first joined Silverchair, AI applications were relatively straightforward—summarization tools, for instance, were among the earliest implementations. However, as AI has evolved, so have the applications. Retrieval-Augmented Generation (RAG) began to emerge, especially with advancements in vector databases and search enhancements. The introduction of Graph RAG offered a new dimension to discovery capabilities, while agentic frameworks have grown in popularity, paving the way for fully agentic systems that incorporate multiple AI methodologies to automate complex workflows.Looking ahead, it’s hard to see any of these applications of gen AI going away. However, as agentic approaches become more common, we’ll likely see these systems integrated in even more innovative ways. There are numerous hybrid approaches to data analysis and knowledge discovery that bring together many of the ideas we’ve already seen, and I believe this mindset will continue to gain traction.
As we continue to build and explore new possibilities, we’re also mindful of not rushing AI into products for the sake of marketing but instead focusing on creating tools that deliver real value to end users. To that end, we’ve dedicated significant time to engaging with clients and end users, both in one-on-one conversations and at conferences, to identify promising directions and eliminate ideas without realistic applications. These dialogues have been incredibly illuminating, revealing both the publishing industry’s enthusiasm for AI and the challenges we must navigate.
The Publishing Industry’s Embrace of AI
One of the most striking takeaways from my experiences at conferences and partner discussions is the publishing industry’s strong receptiveness to AI. Historically, there may have been some hesitation to adopt new technology, but today, we see many publishers actively experimenting with us to bring the next generation of research tools and products to market. This shift in attitude is driven by several factors:- Disruption Awareness: Publishers are acutely aware of the potential disruption AI could bring. Large tech companies have long redirected users away from publisher sites to increase their own advertising revenue and engagement. Publishers now recognize the importance of getting ahead of this with a new wave of user interfaces enabled by conversational tools and visual aids.
- Changing User Behavior: Researchers, medical professionals, and the general public are increasingly becoming accustomed to chat interfaces, large language models (LLMs), and visual tools—and they will soon start to expect them. If the publishing industry lags in adopting these technologies, we risk losing users to competitors who can offer these services. While this transition won’t happen overnight, it’s crucial to lead rather than react.
- Content Generation Risks: The proliferation of LLMs has lowered the barrier to generating research-like content, raising concerns about the potential flood of poor-quality or even forged research. Understanding and addressing these risks is essential to maintaining the integrity of scholarly publishing.
- Efficiency Gains: Publishers are excited about the efficiency gains that AI can deliver through robust systems centered on chat, semantic analysis, and generative capabilities. While AI has been around for years, the ease of integrating it into existing systems and the advancements in machine-level semantic understanding have never been greater. The future holds exciting possibilities for building new tools with these capabilities.
When we first started, our focus was on expanding our knowledge and skills, which proved crucial to our conversations with partners and the broader industry. But while we were learning, we were also building. Since last year, we’ve developed several tools, ranging from early prototypes to those in advanced beta testing. As we’ve discussed and released these tools, the high level of partner engagement has been amazing and truly demonstrates the ongoing interest in AI’s impact.
Looking Ahead: The Future of AI and Publishing
As the initial excitement around generative AI shifts from novelty to practicality, the focus is increasingly on the real value AI can deliver through tangible, proven products. While many promises have been made, AI’s true impact is still unfolding. This gradual evolution aligns well with our team’s approach of continuous learning, experimentation, and long-term thinking. We understand that while innovation drives change, it's the products and tools we develop that ultimately make a difference to everyday users.At Silverchair, we’re seizing the opportunity to carefully and deliberately build tools that will genuinely benefit publishers and researchers. I believe we’re taking the right steps at the right pace to achieve our goals. The changes we’ve made as a company, coupled with the industry’s growing expectations, have significantly influenced my outlook on work and the future of AI in our space.
I’m genuinely excited about what lies ahead for both Silverchair and the publishing industry.