Macmillan Staff Say the Publisher Threw a Party for a Workplace AI Assistant That Was Later Killed

The use of AI by publishers is directly impacting author trust and may lead some to pursue independent publishing routes.

Macmillan, a major book publisher, launched an AI assistant named NOA (No Ordinary Assistant) developed by Chaptr, a company backed by Macmillan's parent organization, Holtzbrinck. Employees were encouraged to use NOA for tasks like composing emails and drafting copy, with assurances of data privacy. However, staff reported NOA was unreliable, prone to "hallucinations," and that usage was restricted to avoid inputting sensitive manuscript data. Despite initial training and encouragement, including using NOA to generate performance review goals, many employees found it unhelpful. In late 2024, Chaptr announced it would discontinue NOA. Macmillan has since begun exploring Google Gemini, though early attempts to use it for evaluating manuscript submissions were deemed "useless" by an editor. Macmillan's communications vice president stated that Gemini is used for business purposes but not for publishing workflows, and disputed the accuracy of NOA's rollout description.

The mixed reception and eventual discontinuation of NOA highlight the challenges publishers face in effectively integrating AI. Despite assurances of proprietary data protection, employee skepticism and the AI's unreliability suggest a disconnect between development promises and practical application in a content-sensitive industry. Macmillan's subsequent pivot to Google Gemini, with its own reported limitations in evaluating submissions, indicates that the search for viable AI solutions is ongoing and fraught with difficulty. This trial-and-error approach suggests that the industry is still in the early, experimental stages of AI adoption.

The broader tension between publishers embracing AI for efficiency and authors' concerns over copyright, creative integrity, and potential job displacement is escalating. Authors are increasingly questioning the value proposition of traditional publishing when AI-driven cost-cutting measures seem to devalue human input and creative judgment. This friction not only strains author-publisher relationships but also fuels a growing consideration among established authors for self-publishing, as they perceive a diminishing need for institutional support if core tasks are outsourced to machines.