The 'Productivity' Lie: Macmillan's AI Party Was a Funeral for Editorial Talent

In January 2024, Macmillan Publishers decided to celebrate the arrival of a new coworker with purple balloons, light-up marquee letters, and a plate of doughnuts. The guest of honor was "NOA," or "No Ordinary Assistant." As Wired first reported, this AI chatbot—developed by the German software company Chaptr and backed by Macmillan's parent company, Holtzbrinck—was pitched as a productivity boon.

But let's call this what it actually was: a corporate performance.

From my vantage point covering the intersection of VC funding and startup hype, the Macmillan disaster is a textbook example of the 'AI productivity' grift. The pitch is always the same: these tools will liberate workers from drudgery. In reality, they are deployed to see how much of the human element can be excised from the P&L before the product collapses.

According to Wired, Macmillan employees were encouraged to use NOA for everything from drafting copy and marketing research to the absurdity of generating their own performance review goals. The tool was so fundamentally disconnected from the reality of publishing that one employee noted a Chaptr representative's claim that they understood publishers deal with "content," prompting the worker to reflect, "And here I thought we were publishing literature."

Despite the festive rollout, the reality on the ground was a failure. Wired reports that employees found NOA unreliable and unhelpful, with many simply refusing to use it. The tool was eventually sunsetted in October 2025. But don't mistake this for a victory for human editors. Instead of returning to a talent-first model, Wired reports that leadership has since attempted to integrate Google Gemini into the publishing process. While Kelly Collins, vice president of communications and community at Macmillan, told Wired there has been no directive to use Gemini for the publishing process—including evaluating submission letters—staffers told Wired they experimented with the tool for that exact purpose and found the results "useless."

This isn't just a Macmillan problem; it's a Big Five contagion. Wired reports that staff at Penguin Random House (PRH) have been encouraged to use ChatGPT Enterprise, with a slide deck from PRH-acquired Hay House labeling the AI a "core productivity, creativity, and collaboration tool." The irony is staggering: while executives push these tools, the German-language division of Penguin Random House and Macmillan are actively suing AI companies for copyright infringement.

When a company sues an AI firm for stealing intellectual property while simultaneously pushing its employees to use that same technology to evaluate manuscripts, it isn't a "thoughtful approach to AI," as Collins describes it to Wired. It is a contradiction.

The goal isn't better books; it's a leaner headcount. By framing AI as a "productivity" tool, leadership creates a metric for failure. When the AI fails—as NOA and Gemini have—it isn't the tool that is viewed as the problem, but the human's inability to "prompt" it correctly.

As Wired reports, bestselling authors are now considering abandoning traditional publishing because of this war over AI. If the Big Five continue to prioritize the efficiency of a chatbot over the intuition of an editor, they won't just be losing their staff—they'll be losing the very authors that make their imprints, like Tor Books and Farrar, Straus and Giroux, relevant. You can't automate literature, but you can certainly automate the death of a publishing house.