AI-enabled audiobooks aren’t for everyone – but their impact is only going to grow.
Last October, Ray Dalio released an AI version of himself. Not a chatbot hawking his books, but a conversational clone trained on decades of his writing, speeches and investment philosophy. Ask it a question about managing your career or understanding markets and it responds in his voice, drawing from his principles. Dalio claims it’s 95% as good as talking to him directly.
New formats are emerging from AI. One I think publishers should watch is the interactive audiobook, where listeners can pause, ask questions and engage in conversation with the content.
Why? Because people are increasingly drawn to talking to content rather than passively consuming it. Google’s NotebookLM went viral last year with a feature that turns any document into a podcast-style conversation between two AI hosts. Millions discovered they preferred discussing content to merely absorbing it. The AI hosts banter with each other and summarise any topic. Crucially, they sound human enough that users forget they’re synthetic.
When publishers talk about AI, they tend to focus on copyright lawsuits and synthetic narration. Both matter. But neither addresses another interesting question: what if books stopped being one-way broadcasts?
Sahil Lavingia, founder of Gumroad, built something similar for his book The Minimalist Entrepreneur back in 2022. Readers can ask questions about the content and receive answers in his voice. It’s crude compared to what’s possible now but it proved the concept: some readers want to talk back.
The technology has caught up. Voice AI latency has dropped to around half a second, not quite the 200 milliseconds of natural human conversation, but close enough that it no longer feels like talking to a call centre. Voice cloning requires 30 minutes of clean audio to produce a professional-quality replica. Retrieval systems can ground AI responses in specific book content, reducing the hallucination problem that makes chatbots unreliable.
Where does this leave publishers?
The audiobook market offers one answer. US sales hit $2.22 billion in 2024, growing 13% annually. More than half of American adults have now listened to an audiobook. Yet the format has remained passive: a lecture you absorb while commuting or doing dishes. Almost half (47%) of percent of listeners admit to getting audiobooks free through YouTube or file-sharing. Attention drifts. Engagement is shallow.
Publishers could change that with interactive audiobooks. A listener could pause a business book to ask how a framework applies to their situation. A self-help title could adapt its advice to the problems someone describes. You could interrogate a historian’s interpretation rather than simply absorbing it.
Fiction is a harder case. Readers often want immersion, and the appeal of a novel lies partly in surrendering control to the author’s pacing. But non-fiction – particularly the practical, applied kind that dominates the self-improvement and business categories – is ripe for transformation.
Interactive formats will not replace traditional audiobooks, just as podcasts never replaced radio. But the success of NotebookLM suggests that some listeners, some of the time, will want to engage differently
Continues...
The obstacles are real. Rights frameworks haven’t caught up; the Authors Guild rightly argues that AI training rights belong to authors, not publishers, and aren’t covered by existing contracts. Voice actors are understandably anxious, having just concluded a strike partly over AI protections. Consumer willingness to try AI-narrated content has actually declined, from 77% to 70%, between 2023 and 2025. And human-narrated audiobooks still generate seven times the royalties of their AI equivalents.
None of this suggests that listeners will stop wanting the passive experience. Many people value audiobooks precisely because they can absorb a story while doing something else. Interactive formats will not replace traditional audiobooks, just as podcasts never replaced radio. But the success of NotebookLM suggests that some listeners, some of the time, will want to engage differently. The question worth asking is whether publishers will shape this emerging format or cede it to platforms and individual authors.
Dalio didn’t need a publisher to build his AI clone. Neither did Lavingia. Google built NotebookLM without industry input. If interactive audiobook experiences become mainstream, publishers risk becoming mere licensors of content that others transform.
The commercial model remains unclear. Who pays for the compute costs of real-time AI conversations? How do you price unlimited access to an author’s expertise? What happens when an AI misrepresents an author’s views? Dalio’s clone reportedly takes a sunnier view of government debt than he does himself.
These are genuine problems.
Publishers are right to defend their copyrights and question the ethics of AI training on unlicensed material. That work matters and should continue. But alongside it, there’s another question which won’t wait: what does AI make newly possible? Formats that didn’t exist before. Relationships between authors and readers that weren’t previously scalable.
The publishers who get there first will have a say in how it develops. The ones who wait may find they’re merely licensing content to those who do.
