Microsoft will soon preview a version of Word that will use artificial intelligence to make your writing not just grammatically but politically correct.
Microsoft doesn’t call it a “political correctness check,” but that’s essentially what it is. Not that there’s anything wrong with that.
Actually Microsoft calls it “Ideas in Word,” which refers to a series of AI-driven features that help you format your document and write better.
For instance, Word will decode acronyms for you, and tell you how long it’ll take to read a given document. It’ll also underline words or phrases that sound insensitive, and suggest corrections.
Say you write, “We need to get some fresh blood in here.” The AI is likely to underline “fresh blood” and suggest “new employees” instead.
It might underline places where your writing exhibited gender bias. If you tend to say “mailman” or Congressman” in the generic, it might suggest you use “mailperson” or “Congressperson.” If you use the term “gentlemen’s agreement,” it may suggest you use “unspoken agreement” instead.
If you describe someone as a “disabled person” the AI would suggest “person with a disability.” Person-first terminology is preferred because it portrays the person as more important than the disability.
The “inclusiveness” checks are part of a larger group of “Refine My Writing” tools that also include clarity, conciseness, punctuation, and “sensitive geopolitical terms.” For that last one, the AI’s models look for phrases that may be hard to understand by, or that might be offensive to, someone in another country or culture, Microsoft says.
We’re a long way from spell check here, folks.
A new kind of PC for Microsoft
Spelling and grammar checks check the user’s words against a fairly agreed-upon set of spellings or usage rules. Correcting words for their “correctness” in the cultural or political sense seems like a more subjective and slippery exercise. Actually, Microsoft hasn’t completely settled on the full list of correctness checks the AI will run on the text.
For the various new checks, Microsoft assembled a team of linguists and other experts to anticipate the poor word choices people might make, and assemble lists of terms that would work better, Office Intelligence product manager Malavika Rewari tells me. The AI’s training data also includes Wikipedia pages, which are constantly being updated and corrected.
The good news is that just as you can ask Word not to give you grammar suggestions, you can go into the settings and tell it not to monitor the correctness or sensitivity of your words.
Personally, I don’t think I would turn off the suggestions, at least not at first. I worry about unknowingly or accidentally inserting terms or references in my writing that convey value judgements that I don’t really mean. Regardless of how I feel about political incorrectness, it must be better to at least know when I’m writing something that might offend. Whether I use the Microsoft AI’s suggestion for improvement is my choice.
Microsoft says it will begin offering a preview of Ideas in Word in June. I’m eager to try it out. Maybe with Word’s new AI superpowers, the review will write itself.