The Case for Transparent AI Use in Writing
A proposed scale reveals how much AI contributes to written content, moving beyond yes-or-no questions.
1The problem with binary questions
Asking writers 'Did you use AI?' oversimplifies the complex ways AI assists in modern writing. From sentence rephrasing to full op-ed generation, AI's role varies dramatically. The case of Stanley Druckenmiller's WSJ piece, scored 100% AI by Pangram despite his claims of partial authorship, highlights this ambiguity. This binary questioning fails to capture the spectrum of AI assistance now common in professional writing.
2Druckenmiller's calculator analogy
The billionaire investor defended his AI use by comparing it to using calculators for math. However, this analogy breaks down because writing lacks single right answers. His unapologetic stance and WSJ's support reflect growing acceptance of AI in publishing. The controversy emerged not from AI use itself, but from the mismatch between his description (partial use) and detection tools' assessment (100% AI).
3The 'Who Wrote This?' scale
Proposed as a solution, this six-level scale ranges from fully human (H5) to complete AI generation (H0). Intermediate levels distinguish research assistance (H4), post-draft polishing (H3), collaborative writing (H2), and AI-first drafting (H1). Similar to The Datatech Times' implementation, it offers nuanced disclosure without requiring perfect human/AI separation that current tools can't reliably provide.
4Protection through transparency
The scale could shield writers from AI witch hunts while maintaining reader trust. By standardizing disclosure, it prevents overreactions to minor AI use like grammar checks. Historical precedents exist - ghostwriting has long been accepted in politics and celebrity memoirs when disclosed. The key is avoiding deception, not banning assistance tools outright.
5Current industry resistance
Despite its benefits, adoption remains rare. Some journalists fear AI detectors' inaccuracies, with WSJ's James Taranto comparing Pangram to a 'defamation machine.' Others modify writing styles (avoiding em dashes) to evade false positives. This defensive posture stems from binary perceptions that the proposed scale could help overcome through graduated disclosure.
6Historical parallels
The debate echoes 1845 accusations against Alexandre Dumas about his 'novel factory.' Then as now, the core issue was attribution transparency rather than collaboration itself. The scale modernizes this centuries-old concern, acknowledging that creative processes often involve multiple contributors while maintaining accountability for final output.
7Implementation challenges
Self-reporting risks dishonesty, but detection tools create counterpressure. Being caught exaggerating human involvement (claiming H4 when actually H0) would damage credibility more than honest disclosure. The system relies on professional norms developing around appropriate AI use levels for different content types, similar to existing ghostwriting standards.
8Toward ethical AI collaboration
The proposal shifts focus from prohibition to responsible use guidelines. As AI becomes writing's 'calculator,' the industry needs frameworks distinguishing ethical assistance from deceptive automation. Standardized disclosure preserves human authorship's value while embracing tools that enhance productivity, provided their role is transparent to readers.
Original article: Fortune | FORTUNE — The Case for Transparent AI Use in Writing