Case File: When AI Meets the Pen: A Plain-English Dissection of the Boston Globe’s ‘AI Is Destroying Good Writing’ Claim

Background: The Opinion Piece That Sparked the Debate

In a cramped newsroom on a rainy Boston afternoon, a senior editor stared at a screen flashing the headline, "AI is destroying good writing." The Boston Globe’s op-ed, published last year, framed the controversy as a cultural battle between algorithmic speed and human craft. Pegasus in the Shadows: How the CIA’s Deception...

The article quoted a veteran columnist who warned that "the flood of machine-generated prose is eroding the discipline that once defined journalism." That line alone set off a cascade of reactions across corporate boardrooms, where managers worry about the trade-off between productivity and brand voice.

According to the Globe, the piece was intended as a wake-up call, not a manifesto. It cited a recent study showing that AI-assisted drafts reduced editing time by 30 percent, yet the same study noted a 12-point dip in reader satisfaction scores.

The Core Challenge: Balancing Speed with Substance for Non-Technical Leaders

Non-technical managers often hear two competing narratives. On one side, AI promises to churn out reports in minutes, freeing staff for strategic work. On the other, the Globe’s op-ed warns that such shortcuts can produce bland, formulaic copy that fails to engage audiences.

For a mid-size tech firm, the dilemma manifested in a quarterly newsletter that dropped from a 4.2-star rating to 3.5 after an AI-generated rewrite. The drop was quantified by a post-campaign survey, which recorded a 15-percent increase in unsubscribe rates.

Complicating matters, a separate Boston Globe report highlighted that students at Berklee College of Music were paying up to $85,000 for AI classes that many deemed a waste of money. The parallel suggests that even costly training does not guarantee better outcomes, a cautionary note for managers budgeting AI tools.

The Practical Approach: A Six-Month Pilot in a Global Marketing Team

To test the Globe’s warning, a multinational consumer-goods company launched a controlled pilot. The team split into two groups: one used a leading generative-AI platform for drafting press releases, the other relied on traditional copywriters.

Both groups followed identical briefs, timelines, and approval chains. The AI group received a brief on day one, generated a first draft within four hours, and then spent two hours on human refinement. The human-only group took an average of 18 hours to produce a comparable draft.

Key performance indicators (KPIs) included turnaround time, readability scores (Flesch-Kincaid), and stakeholder satisfaction. The AI-augmented drafts averaged a readability score of 68, slightly higher than the human-only average of 64, indicating marginally easier comprehension.

The pilot revealed that AI can shave 70 percent off raw drafting time while maintaining comparable readability.

However, the same KPI sheet showed a 9-point gap in stakeholder satisfaction, measured on a 100-point scale, favoring the human-only group. Managers noted that the AI drafts often lacked the nuanced brand tone that seasoned writers inject.

Measured Results: Numbers That Tell the Full Story

At the end of the six-month period, the company compiled a comprehensive report. The AI-enabled workflow produced 312 press releases, versus 124 from the human-only stream, a 152-percent increase in volume.

Despite the volume boost, external media pickups for AI-generated releases were 22 percent lower than for human-crafted pieces. The Globe’s op-ed highlighted a similar trend, noting that "quantity does not equal quality when audience trust is at stake."

"The flood of machine-generated prose is eroding the discipline that once defined journalism," - Boston Globe opinion column, 2023.

Financially, the AI stream saved the company an estimated $210,000 in labor costs, based on an internal rate of $75 per hour for senior writers. Yet the lower engagement translated into a $95,000 shortfall in earned media value, as calculated by the firm’s media analytics team.

When the numbers are juxtaposed, the net gain shrinks to $115,000 - a modest improvement that masks the qualitative concerns raised by the Globe.

Lessons Learned: What the Pilot Teaches About AI and Writing

First, speed does not automatically translate into impact. The Globe’s warning about “destroying good writing” finds empirical support in the lower stakeholder satisfaction scores, even when readability improves.

Second, human oversight remains indispensable. The pilot’s two-hour refinement phase proved critical; without it, the AI drafts fell further behind on brand consistency.

Third, investment in AI training must be strategic. The Berklee tuition figure - $85,000 for AI courses - illustrates that pouring money into education does not guarantee ROI unless the curriculum aligns with real-world editorial needs.

Finally, metrics matter. The company’s decision to track both quantitative (turnaround time, cost) and qualitative (satisfaction, tone) KPIs prevented a one-dimensional view of success.

These insights echo the Globe’s call for a balanced approach: embrace technology, but safeguard the craft that gives writing its persuasive power.

What We Can Learn: Actionable Takeaways for Non-Technical Managers

Non-technical leaders should treat AI as a collaborative tool, not a replacement. Deploy AI to generate first drafts, but allocate dedicated time for human editors to infuse voice and context.

Set clear performance benchmarks that capture both efficiency gains and audience response. A 30-percent reduction in drafting time is impressive, but a simultaneous dip in satisfaction signals a hidden cost.

Invest in targeted AI training that mirrors actual workflow challenges, rather than generic courses that may cost upwards of $85,000 without delivering relevant skills.

Finally, maintain a feedback loop. Regularly survey internal stakeholders and external audiences to ensure that the speed AI provides does not erode the credibility that good writing builds.