Grammarly's 'Expert Review': AI Necromancy Without the Consent
# Grammarly's 'Expert Review': AI Necromancy Without the Consent
Grammarly's latest gimmick, Expert Review, sounds like a writer's dream: an AI agent that channels the spirits of great minds to polish your prose. Drop in your draft, and it summons feedback from domain experts—alive, dead, or somewhere in between—tailored to your topic. Need astrophysics tweaks? Hello, Neil deGrasse Tyson. Medieval history? Maybe the late David Abulafia, who passed in January 2026. Users rave: one DEIB director calls it "trustworthy expert eyes," while a researcher digs the mentor-like tips with examples.
<> But here's the rub: this isn't expertise—it's digital identity theft./>
The Ghoulish Underbelly
Critics aren't mincing words. Historians like Dr. Verena Krebs are horrified, accusing Grammarly of 'necromancy' by resurrecting deceased scholars via scraped public works, no consent required. Associate professor Vanessa Heggie blasts it as "obscene," with Grammarly crafting mini-LLMs or persona prompts from academics' life's work to mimic their voice. Even living celebs like Tyson get dragooned without a nod. It's not innovation; it's escalation. As one observer quipped, "At least they're not hiding it anymore."
Grammarly (fresh off rebranding to Superhuman, because why not?) markets this as "insights from subject-matter experts and trusted sources." You can customize experts, tweak for your niche, and watch the AI evolve with your draft. Tech-wise, it's slick—proactive suggestions on clarity, tone, engagement, baked into their spell-check empire. But ethically? A dumpster fire. This mirrors the AI industry's dirty secret: hoover up human creativity, regurgitate as 'new,' and cry 'public domain' when called out.
Why Developers Should Care (and Boycott)
As devs, we're building the future—but not like this. Grammarly's approach normalizes scraping without permission, eroding trust in AI tools. Imagine your code reviewed by a 'ghost' of Linus Torvalds, trained on your repos sans ask. Creepy? Unethical? Both.
- Consent is king: No permission means no feature. Full stop.
- Transparency fail: They flaunt names but hide the training sausage-making.
- Legal landmine: Lawsuits loom as academics push back—expect class actions.
- Better alternatives: Open-source LLMs with ethical datasets beat this necromantic nonsense.
Grammarly boasts ubiquity and proactivity in their 2026 AI manifesto, embedding agents everywhere. Fine, but not on stolen backs. Positive user buzz can't whitewash the foundation: a house of pilfered cards.
The Verdict: Skip the Specters
Expert Review isn't elevating writing; it's cheapening legacies. Grammarly, retract, compensate, and rebuild with consent—or watch users flee to tools that respect creators. Devs, demand better: code AI that innovates, not impersonates. Your next pull request deserves real experts, not AI zombies.
(Word count: 512)
