Publishing Exec: AI Literacy Means Understanding Systems, Not Just Tools

Publishing professionals increasingly use AI tools daily, from chatbots to translation assistants. Yet few understand how these systems actually work—a knowledge gap that industry consultant Javier Celaya calls "one of the most consequential blind spots" facing publishers today.

Celaya, founder of cultural sector consultancy Dosdoce.com, enrolled this year in an executive master's program in artificial intelligence at Spain's Instituto de Inteligencia Artificial. His goal wasn't learning to write better prompts, but gaining rigorous understanding of how models are built, trained, and governed—and where their genuine limits lie.

Why it matters

Publishers face strategic decisions about automation, rights management, and partnerships with AI companies. Leaders who understand AI at a structural level—not just as end users—can evaluate vendor claims, design appropriate governance, and identify growth opportunities beyond cost reduction. Surface-level tool familiarity won't suffice for these decisions.

From Tool User to Informed Skeptic

The program fundamentally changed how Celaya approaches AI, he writes in Publishing Perspectives. Before opening any tool now, he thinks specifically about what he wants an AI agent to accomplish. When vendors claim their model uses "clean data" or sidestep questions about data ownership, expiration dates, or reuse policies, he knows which questions to ask.

Particularly valuable was studying AI adoption across healthcare, automotive, pharmaceutical, and entertainment sectors. Examining where other industries stumbled, which use cases delivered value, and what governance structures emerged provided comparative context publishing lacks. "In publishing, we tend to talk about AI from a defensive position, while the rest of the world has already kicked off projects using innovative tools that we are still debating," Celaya notes.

Beyond Cost Cutting to Growth Strategy

Publishing's AI focus remains stuck on cost savings—deploying automation to trim expenses. These applications represent the floor of AI's potential, not the ceiling, according to Celaya.

The strategic questions defining the next decade are barely being asked: Which publishing processes can be fully automated? Which require humans at workflow endpoints? Which must remain entirely human? Most critically—how can AI enable growth rather than simply shrinking cost bases?

AI-powered translation and production could bring key titles to entirely new audiences at previously impossible scale and speed. Marketing and distribution could shift from local to genuinely global when language and localization are no longer bottlenecks. Yet few publishers are analyzing these possibilities with appropriate depth.

Partners and Adversaries Simultaneously

The industry's most difficult question concerns its relationship with AI companies. Celaya's answer after serious reflection: both partners and adversaries.

Many AI companies built their capabilities by training on copyrighted works without consent or compensation. The legal and moral case for redress is strong, and publishers should pursue it. But lawsuits and business partnerships aren't contradictions—they're parallel tracks that can and should run simultaneously.

Publishers and AI companies should jointly define collaboration scenarios and reimagine distribution channels that will shape how content is discovered and consumed.

Education as Strategic Requirement

Celaya goes beyond recommending continued AI education—he argues it should be required for anyone in management. No managing director can responsibly define AI strategy without understanding at a structural level what AI can and cannot do.

AI isn't another incremental shift. It touches every publishing function: creation, curation, production, translation, marketing, distribution, and rights management. Leaders who navigate this era most successfully will understand AI deeply enough to ask the right questions of their teams, technology partners, and the models themselves.

These details were first reported by Javier Celaya in Publishing Perspectives.