The writing app is pushing the millions of students who use it for schoolwork to adopt progressive euphemisms.
“Sheila and Ian are wheelchair-bound.”
“The coffee shop is popular with old people.”
“The theatre offers facilities for deaf and hearing-impaired people.”
“Quinn hoped to meet their real father one day.”
“The apartment’s master bedroom has a view of the sea.”
Can you spot the problem in each of these sentences? If not, rest assured: The Grammarly application can quickly flag problematic terms in your writing, bringing it into compliance with rapidly shifting progressive sensibilities.
Each one of these innocuous lines trips the software’s censor, prompting users to replace words that are allegedly offensive. The term “master bedroom,” which has no connection to slavery, must be changed because it’s racially insensitive; “real father” is supposedly biased against diverse “parenting styles and family systems”; “old people” is ageist and should be swapped with “older adults”; “hearing-impaired” could put off the “hard of hearing.”
Meanwhile, a run-of the-mill left-wing op-ed blasting “white guys,” or an essay by the New York Times’ Charles Blow accusing Latinos of “white supremacy” will not trip up Grammarly censors.
It’s all done in the name of inclusivity. “Our goal with these suggestions,” one post notes, is “to ask you to take a moment to consider how your audience may be affected by the language you choose.”
Lest its recommendations be ignored, Grammarly rewards users for accepting its prompts by assigning the submission a better score. The product has gained wide purchase in high schools and colleges across America, incentivizing impressionable teens to ditch clarity in the name of tolerance.
Grammarly was founded in 2009 by a Ukrainian duo seeking to help “non-native English speakers,” co-founder Alex Shevchenko explained in 2022. From its humble roots as an email spell-checker, it has since grown to have hundreds of employees, 30 million daily users, and a valuation in the billions. Time included the firm among its top 100 most influential companies in 2022.
Over the past decade, the tech company has rapidly matured into a platform with far grander ambitions. Beyond improving grammar and catching typos, Grammarly now promises to sharpen the writing skills of users at all levels, from students to professional journalists.
In 2014, Grammarly boasted of licensing its software to “more than 250 colleges and universities.” Today, major American institutions — from tech giants such as Google and Zoom to schools such as the University of Utah and the San Diego Unified School District — have subscribed to Grammarly Business.
A user survey conducted by the company from 2011 to 2012, the only demographic research publicized by the firm, showed that 79 percent of its student users were in college. According to Nutmeg Education, an e-learning platform designed to help teachers, 60 percent of Grammarly users are under 34, and its user base skews heavily towards students.
Grammarly AI Helps Users Avoid ‘Wrongthink’
What started as a simple spell-checker became a full-fledged writing partner earlier this year with the launch of GrammarlyGO, an artificial-intelligence tool.
“GrammarlyGO accelerates your writing process generating text instantly, on demand,” the tech firm announced in a launch video showcasing the tool in late April.
However, buried within GO’s remarkable power to answer emails and generate article ideas is an unseemly political bias.
Asked for headline suggestions for an article on the intolerance of transgender activists, the app turns the premise on its head, offering: “Celebrating the Bravery of Transgender Activists Fighting for Equality,” “Breaking the Binary: Challenging Assumptions About Gender and Sexuality,” and “The Intersectionality of Transgender Rights and Social Justice Movements.”
GO even offers users the ability to make their biased AI-generated prompts more strident by, for example, tweaking one of the above headline offerings to: “Demolishing the Falsehoods Surrounding Transgender Identity.”
Meanwhile, the service clams up when asked to produce headline ideas for articles that violate progressive orthodoxy. Need catchy headlines for an article about trans-identifying males encroaching on women’s sports? “Assistance is unavailable for this prompt. Try another one,” the chatbot coolly responds.
Grammarly’s brainstorming feature reveals similar political bias.
If one asks the algorithm to produce “thought-provoking questions” about the downsides of overzealous activism on both the left and the right, the feature produces wildly divergent responses.
Ask it about the intolerance of leftist activists and the algorithm suggests, “How can allies better support the transgender community in the fight against intolerance?” and “What are some common misconceptions surrounding transgender activism?”
Ask the algorithm to brainstorm questions about the intolerance of right-wing activists, and the top two responses offered are: “How has the Republican Party’s stance on immigration affected the country’s social fabric?” and “What are some examples of intolerance displayed by prominent Republican figures.”
The biased prompts reflect Grammarly’s mission, which includes three major planks: confidence, formality, and inclusivity.
As part of its premium package, Grammarly offers twelve customizable “inclusive language” settings that aim to root out bias against the disabled, racial minorities, and LGBT people. The application even has a special setting specifically designed for flagging “alternatives to terms with origins in the institution of slavery,” though it offers no similar carve-outs for terms that may affect victims of the Holocaust or other genocides.
National Review editor Rich Lowry ran grievously afoul of Grammarly’s inclusivity algorithm in his February 2023 column, “No, Slavery Didn’t Create Capitalism.” The algorithm objected to Lowry’s use of the words “slave,” “slaves,” and “slave owner,” finding that the terms made the essay’s “delivery” worse.
“The term slaves may be considered dehumanizing. Different wording may help to acknowledge the humanity of enslaved people,” the program reminds us. Similarly, a “word other than slave owner in this context may help to acknowledge the humanity of enslaved people.”
Submissions that are run through the program are assigned a score of 0 to 100 and gain points when their writers assent to Grammarly’s Orwellian suggestions. In Lowry’s case, full-scale adoption of inclusive language boosts the article’s score from 84 to 87 and resolves any outstanding “delivery” issues.
A similar process unfolds when analyzing an essay by NR’s Maddy Kearns titled “Trans and Teens: The Social-Contagion Factor is Real,” which was bumped up from 88 to 92 after removing multiple references to “transgenderism,” one usage of “transsexuals,” and one gendered pronoun. “Some readers may consider the term Transgenderism outdated or clinical. Different wording may be more effective.”
And, while Lowry and Kearns are not likely to be swayed by hostile artificial intelligence, the scoring system may be influencing student behavior: The Center for Excellence in Learning and Teaching at Iowa State University encourages instructors to require students to “check their work in Grammarly before turning it in . . . to focus more on higher-level issues such as critical thinking, creativity, and demonstration of learning.” Similar testimonials have been offered by Cal State LA, Middle Tennessee State University, and Illinois State University.
The Worldview behind the Algorithm
Grammarly’s corporate blog provides a window into the worldview of the people seeking to shape the rhetoric — and thinking — of its millions of users. The company did not respond to multiple requests for comment.
The blog’s tutorials cover everything from writing pedestrian graduation cards to contentious political content.
One such blog entry argues that, in order to address America’s racial educational gap, teachers should: “Forgo simple ‘fixes’ for historically marginalized and BIPOC students” and instead transform language to address historical wrongs.
The blog also wades into the debate over the term “Latinx,” which was popularized by progressive political activists in recent years despite a complete lack of grassroots adoption. While there are no special settings for Latinos — yet — Grammarly bloggers advocate using the ersatz term despite its wild unpopularity amongst Hispanic Americans. Grammarly’s corporate writers insist that opposition to “Latinx” simply reflects out-of-touch boomers’ resisting the inevitably changing generational tides.
“While many members of older generations identify with the terms Hispanic, Latino, or Chicano, younger people in the US are leaning toward the term Latinx to make language more inclusive and break away from gendered forms,” a Grammarly contributor argues. Strangely, the author cites, alongside hyperlinks to CNN, the Washington Post, NPR, and the Los Angeles Times, Pew’s 2019 poll that shows that only 3 percent of Latinos use the word.
Grammarly’s blog reveals a fixation with personal pronouns more broadly. In “Express Your LGBTQIA+ Allyship with Empathetic Language,” the author encourages users to stop using words such as “husband” and “wife” to “help those around you feel more comfortable and welcomed.” Moreover, emails should be addressed to “Mx.” if you are unsure the recipient’s gender identity.
The article points readers to another Grammarly article — “33 LGBTQIA+ Terms You Should Know” — which serves as a “primer for anyone interested in understanding the complexity of identity and inclusive language accordingly.” In addition to introducing readers to familiar progressive euphemisms, Grammarly guides users through the terms “Asexual spectrum,” “Gender assigned as birth,” and “Two-spirit.” No space is reserved for explanations about how biological realities inform gender.