Stuart Leitch

Artificial intelligence and large language models have the potential to change, even disrupt, every aspect of scholarly publishing, from how infrastructure and platforms are developed to how content is discovered, used, and licensed. This keynote offered a conceptual overview of the technologies and possibilities.

It has been said that software ate the world. Now, AI is coming to eat software. AI models are becoming disproportionately stronger in domains with easily verifiable solutions, such as mathematics and formal reasoning. Software programming is likewise a bounded domain with strong reward signals for models to self-learn. It’s now possible to generate entire toy applications from a single prompt. The very nature of software development is about to get turned on its ear and it’s likely that at progressive companies, humans will be writing only a small fraction of code commits within 5 years. It’s early days in this rapid transformation, but the direction of travel is clear. On-platform, AI functionality is drastically changing the user experience through workflow improvements, improved discovery, and added capabilities. Finally, we're seeing disintermediation trends in traffic patterns as humans increasingly use agentic AI to search and synthesize primary literature. In other words, AI is changing platforms from both ends—where does that leave us?

Open source or proprietary technology? Community owned or commercial infrastructure and solutions? Can we create competitive, differentiating features or is all publishing technology a commodity? How do you weigh ownership challenges versus partnership risks? These big questions underlie many of our organizations’ most important business strategy and technology decisions. How are some of our industry’s technology leaders grappling with these issues?

Software development is a complex field with a high project failure rate. This presentation surfaces important considerations that publishers should factor into their decision making and management practices concerning software development.