Sociotechnical systems in software and AI: Ghostwriting has always existed — why does AI ghostwriting feel different?

Ghostwriting is not new, but large language models change the sociotechnical landscape around writing. This post unpacks why AI-assisted ghostwriting feels different, what remains uniquely human, and pragmatic guidance for practitioners who build, commission or partner with ghostwriters and writing tools.

Introduction: the ghost on the page

Ghostwriting has always lived in a grey area. Professional writers produce words that someone else signs off on; editors shape voice; friends and assistants tidy memoirs into publishable narratives. These practices predate AI by decades. Yet when a large language model generates a draft in seconds, many people sense a shift that goes beyond speed. That shift is sociotechnical: it’s about how people, organisations and technical systems reconfigure the work of making meaning.

Why old ghosts and new ghosts feel different

At its core, traditional ghostwriting is a social collaboration. A human ghostwriter listens, interviews, iterates and learns to capture a client’s voice. That relationship—sometimes almost therapeutic—creates trust and produces material that feels authentic because it is anchored in human exchange. As The Atlantic reported, the best outcomes come from close collaboration, mutual respect and the writer acting as a trusted adviser rather than only a text generator.

AI changes the sociotechnical assembly that produces text. Language models aren’t another human collaborator; they are tools that alter who contributes what, how quickly, and with what traces of provenance. As researchers at Translated’s Imminent note, these systems reconstruct the relationship between language and calculation and, with it, our ideas about authorship and meaning. That conceptual shift — words being produced by a calculation rather than a conversation — is partly why AI ghostwriting feels qualitatively different.

Three practical differences that matter to practitioners

Voice capturing is harder to outsource to a model. Experienced ghostwriters repeatedly emphasise voice as the centrepiece of the craft. AI can mimic register and patterns, but many professionals find model outputs bland, clichéd or full of jargon—the kind of language they intentionally avoid. Practitioners still rely on human craft to shape nuance and deliver authenticity.
Speed alters process and expectations. AI can draft at scale and speed, collapsing weeks of work into hours. That can be liberating, but it also changes negotiation dynamics: clients may expect faster delivery or more drafts for the same fee, and teams may be tempted to skip essential interviews and edits that produce depth.
Traceability and ethics become design problems. Old ghostwriting already raised ethical questions about attribution and consent. AI introduces technical opacity and reproducibility concerns: which prompts, data or pre-existing texts influenced the output? Managing those questions becomes part of the sociotechnical design of any project that uses AI.

Where AI helps and where it doesn’t

Practitioners who use AI often treat it like a specialised instrument rather than a replacement. Panels at recent industry gatherings emphasised that AI is useful for prepping interviews, doing background research, and producing first drafts that are then humanely sculpted. One practical tip that surfaced repeatedly is to use AI for client due diligence: it’s “cheating, but really good cheating” to gather facts before you meet a potential client.

Conversely, many ghostwriters refuse to let models write core content. They report that AI drafts tend to be generic and miss lived experience. The Atlantic and other commentators note that the human dimension—storytelling, interpretation, emotional intelligence and the ability to be a thoughtful sounding board—remains hard for models to replicate. In short: AI can accelerate the mechanics of writing, but the deeper collaboration and craft stay human-led.

Practical guidance: integrating AI into ghostwriting responsibly

If you’re a ghostwriter, commissioning editor, product manager building writing tools, or an author considering AI help, here are pragmatic steps to keep quality high and relationships clear.

Define roles up front. Be explicit about what the AI will do (research, draft, polish) and what humans will do (voice capture, interviews, final edits). Put this in the contract and client conversations.
Preserve the interview process. Don’t skip human conversations. Interviews are where nuance, anecdotes and the author’s cadence appear. Use AI to summarise interviews, not to replace them.
Create a style and voice guide. Before you hand anything to a model, codify key phrases, taboo words (the “word jail”), tone rules and examples of authentic passages. This reduces the time spent rescuing bland drafts.
Use AI for research and drafts, but keep humans in the loop for narrative decisions. Let models gather facts, propose structures or draft scene sketches. Reserve final decisions about emphasis, ethics and memoiral truth for human collaborators.
Document provenance and consent. Track which sections were model-generated and make sure clients consent to that approach. This protects both parties if questions about originality or source material arise.
Price for human labour. If you use AI to speed drafting, make sure contracts reflect the continuing need for interviews, revisions and editorial judgement. Cheap AI output does not eliminate the value of human insight.
Invest in skills. Writers should learn prompt craft, model limitations and post-editing techniques; product teams should design interfaces that make provenance and human oversight obvious.

Design implications for software and teams

From a sociotechnical perspective, introducing AI into ghostwriting is less about replacing people and more about reconfiguring workflows and responsibilities. Teams building authoring tools should design for transparency, easy handovers between model and human, and audit trails that capture who changed what and why. Small process changes—like mandatory interview notes or a sign-off step on model-generated sections—can preserve quality and trust.

Conclusion: the ghost remains, but the haunt changes

Ghostwriting has never been purely technical or purely social. AI changes the machinery of production and forces us to rethink authorship, traceability and practice. But the qualities that make a ghostwriter valuable—deep listening, editorial judgement and the ability to elicit and shape human experience—haven’t been automated away. For practitioners, the pragmatic path is to treat AI as a powerful instrument and to redesign agreements, workflows and interfaces so that human craft continues to anchor the work.