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

Ghostwriting isn't new, but AI-driven drafting upends familiar sociotechnical arrangements. This post uses a sociotechnical lens to explain why AI ghostwriting feels different and offers practical guidance for writers, editors and product teams navigating this change.

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

Ghostwriting is not a modern invention. As historians and journalists note, the practice stretches back more than a century and has long played a role in publishing, celebrity memoirs and professional communication. Yet when an AI produces prose that someone else stakes a claim to, the reaction is often sharper, more immediate and more morally agitated than when a human ghost does the same work. Why?

To understand the dissonance, it's useful to think in terms of sociotechnical systems: artefacts (models, code, data), human actors (authors, ghostwriters, editors), organisational practices (contracts, editorial workflows), and wider social norms (credit, reputation, labour expectations) all interact. An AI is not simply a new kind of pen; it reshapes the relations among those components. Below I unpack the differences, ground the discussion in recent reporting and commentary, and offer practical guidance for anyone working with words or building systems that produce them.

What’s the same: ghostwriting’s familiar contours

When you strip away the technology, there are substantial continuities between traditional ghostwriting and AI-assisted drafting. Professional ghostwriters emphasise voice capture, iteration and collaboration: they interview clients, draft language that reflects the client's expertise, and often hand over text that the named author edits or approves. Commentators have pointed out that this outcome — somebody else’s words under another person’s name — is the common denominator that makes both human and AI ghostwriting ethically ambiguous in some contexts.

That continuity matters because it reminds us that the ethical questions are not entirely new. As recent coverage shows, ghostwriting has been part of literary and professional life for a long time; it is embedded in publishing practices, power relations and commercial arrangements. The sociotechnical perspective encourages us to treat those arrangements as changeable rather than immutable.

What’s different: scale, speed, opacity and changing labour relations

Several features of AI reshape the sociotechnical landscape in ways that feel different to people who have previously accepted human ghostwriting.

  • Scale and speed: An AI can draft, re-draft and iterate in seconds or minutes. That amplifies production and makes ghostwritten content ubiquitous in ways human labour cannot economically match.
  • Opacity: Many models are black boxes to end users. Human ghostwriting usually leaves a social trail — contracts, emails, edits — whereas AI can produce fluent text without clear provenance, which unsettles norms around authorship.
  • Low marginal cost: Using AI to produce drafts dramatically lowers the cost of producing text. Where a human ghostwriter charges significant fees, AI changes the economics for individuals and organisations.
  • Shifts in craft: Even when AI is used only to polish, commentators observe a subtle erosion of the craft of drafting. Many writers now use models for final edits, which changes how skill is practised and transmitted.

These differences help explain the stronger emotional reaction to AI. It’s not merely that the words come from a machine; it’s that the machine reconfigures how humans relate to the work, to each other and to institutions.

Evidence from practitioners and events

Conversations among professional ghostwriters, journalists and commentators illustrate these dynamics. Practitioners stress that capturing an author’s voice remains the core task; some find AI useful for research and preparation, but refuse to let it write the actual content because its output can be bland and full of clichés. Others use AI for fact-gathering and client due diligence — what one commentator called “cheating, but really good cheating.” These accounts highlight that AI tends to be adopted into existing workflows in pragmatic, uneven ways.

At the same time, observers note that many pieces of contemporary writing already receive AI-based polishing. This widespread use of tooling for final edits is changing expectations about what counts as human writing and what counts as acceptable assistance. The history of ghostwriting shows that the practice has long adapted to new tools and norms; the difference now is the pace and the technological affordances.

Practical guidance: working with AI while retaining trust and craft

For writers, editors and product teams, the sociotechnical framing suggests that the solution is not simply to ban or endorse AI, but to design practices that align technical affordances with social expectations. Here are pragmatic steps you can take.

  • Be explicit about roles and provenance. Create simple ways to record whether AI was used in research, drafting or polishing. Transparency reduces uncertainty and preserves trust.
  • Preserve voice through human-in-the-loop workflows. If a named author must sound like themselves, make human editing a required step. Use tools for research and structure, but let the human shape phrasing, anecdotes and nuance.
  • Use AI where it adds clear value. Employ models for client due diligence, background research or for generating outlines and alternative phrasings. Practitioners report that AI is especially useful for these preparatory tasks.
  • Guard the craft. Practise drafting without AI occasionally. Many writers feel that relying too much on automated polishing diminishes their ability to produce original draftwork; deliberately keeping some processes human helps retain skill.
  • Set organisational policies. Legal and editorial teams should spell out when AI use is permitted, how attribution is handled, and what contracts with ghostwriters should specify about tools and deliverables.
  • Design for auditability. Where feasible, keep logs of prompts, sources and model outputs. Sociotechnical systems become easier to govern when technical artefacts co-exist with records that support accountability.

When to lean into AI, and when to be cautious

There are contexts where AI assistance is an unalloyed good: early-stage research, creating multiple outline options, or doing background checks on subject matter. Conversely, high-stakes contexts — legal testimony, academic authorship, undisclosed political messaging — demand stricter scrutiny and often human-only workflows or explicit disclosure.

Remember: the question isn’t whether AI can write; it’s what social practices we want to surround that capability. If we design policies and workflows that preserve voice, provenance and craft, AI becomes a feature of the sociotechnical system rather than a mysterious replacement for human labour.

Conclusion

AI has not invented ghostwriting; it has reconfigured it. The technology magnifies certain features (speed, scale, opacity) and exposes latent tensions around authorship, labour and credibility. A sociotechnical approach invites practitioners to redesign the relations among people, tools and institutions so that transparency, stewardship and craft survive — and perhaps even thrive. That requires candid conversations, clear policies and a willingness to keep humans in the loop where it matters most. Also, the occasional draft written the old-fashioned way can do wonders for morale.

Sources