Publishers Need “Humans in the Workflow”
“Human in the Loop” is one of the most over-used phrases in the AI conversation. It is meant to reassure us that AI will not operate entirely on its own, especially when important decisions are involved. The machine may recommend or predict, but a person remains involved.
That is a sensible principle. But it is not enough.
The real question is not simply whether a human reviewed the AI output. The more important question is whether that human/AI combination had all the right information, authority, and process to make meaningful decisions. AI can miss nuance, and a human may approve an AI-generated suggestion and still miss the larger issue if the relevant data like rights, royalties, customer history, title status, or financial impact are hidden somewhere else in the organization.
In other words, a human is not truly “in the loop” just because they are asked to click “Approve.” In publishing, the human needs to be in the workflow.
The Limits of a Comforting Phrase
“Human in the Loop” is useful, but it is too vague. It can mean different things in different organizations.
A more useful distinction is to think in terms of three levels. First, there is the human in the loop, where AI assists but a person must approve the action. Second, there is the human on the loop, where AI or automation runs a process and people manage exceptions. Third, there is the human in command, where people define the rules, permissions, governance, and acceptable uses of AI before the system is even deployed.
Publishers need all three. The business is too complex, and too relationship-driven to rely on AI output without meaningful human judgment. And that judgment must be supported by transparent, connected data and clear process design.
Why Publishing Is Different
Publishing is not a generic content business. It is built on unstructured intellectual property and human relationships. Simple decisions can have consequences across the organization.
Take metadata, for example. AI can suggest keywords, summaries, and BISAC categories that save time and improve consistency. But someone still needs to verify that the description is accurate, the positioning fits the list, and the metadata does not misrepresent the work.
Both publishing judgment and AI depend on access to reliable transactional information.
AI Does Not Fix Broken Workflows
AI cannot magically overcome weak operational foundations. If the underlying data is fragmented, outdated, inconsistent, or incomplete, AI will simply accelerate confusion.
Ask an AI assistant whether Spanish-language audiobook rights are available in Latin America, and the answer depends on the quality of the rights data. Ask which authors are affected by a royalty adjustment, and the answer depends on contract and royalty records. Ask which accounts are eligible for credit on a returned title, and the answer depends on customer terms, title status, return rules, order history, inventory, and finance.
The usefulness of AI clearly depends on the quality of the system of record.
That is why “Human in the Loop” should not be treated as a slogan. It should be treated as a minimal operating principle. The goal is to design publishing workflows in which AI, data, and human responsibility are connected from the start.
From Review to Responsibility
For publishers, the better concept may be “Human in the Workflow.”
In that model, AI supports the work but does not own the decision. The editor still owns editorial quality and positioning. The rights team still owns licensing judgment. The royalty manager still owns exceptions and author payment accuracy.
The value of AI is not that it removes people from the process. The value is that it helps people see more, understand faster, and make better decisions.
Workflow Accelerators
knkMedia’s perspective is relevant. knkMedia is built around the idea that publishing processes should be connected end to end. From product concept through finance, data should not live in disconnected silos.
The most valuable AI use cases in publishing will not be isolated tricks. They will be workflow accelerators where AI can see patterns in integrated data that humans may not and help teams understand where focus is required.
Those use cases only become truly valuable when they operate against reliable business data and within controlled workflows. Publishers need more than answers from AI. They need to know where the answer came from, which data supports it, who reviewed it, and what downstream processes are affected.
The Future Is Human Judgment, Better Supported
The AI conversation swings between excitement and anxiety. Some believe AI will transform everything for the better almost immediately. Others see only doom and gloom.
AI is now part of the operating fabric of the business. It helps teams search, summarize, classify, forecast, and analyze. It reduces repetitive work and makes complex questions answerable.
The winners will be the ones that understand where AI belongs, where human judgment must remain central, and how both should be connected inside disciplined workflows.
Human in the Loop is a useful beginning.
Human in the Workflow is the real standard.