It is my understanding that AI usage must be disclosed when used for articles in the Globe and Mail. I fear that may not have happened in this case, and that there are strong indicators it should have.
Dear Sandra E. Martin and David Walmsley,
I am writing on behalf of Canadians for Justice and Peace in the Middle East (CJPME) to express concern regarding the recent op-ed by Irwin Cotler, Mark L. Berlin and Alan H. Kessel, "The Canadian Museum for Human Rights has failed its mandate."
It is my understanding that AI usage must be disclosed when used for articles in the Globe and Mail. I fear that may not have happened in this case, and that there are strong indicators it should have.
As a regular reader of the Globe's Opinion section, I was struck by the mechanical cadence of the prose in this article. It felt stilted in a way that AI generated text is notorious for.
The piece features numerous sentences that I felt contained the hallmarks of AI-generated text. Even the opening lines use a common AI framing: "The Canadian Museum for Human Rights is about to do something no national museum should do: present a deeply contested political narrative as settled historical truth." Emphasis on significance, framed in this exact manner, is standard for AI writing.
As such, I decided to run the entire piece through two reputable AI detection tools. Pangram and Originality.ai both flagged the entire article as 100% AI generated with a high degree of confidence.
I reached out to Max Spero, a former Google machine learning engineer and the current CEO of Pangram, the AI detection platform, and asked if the AI detection results seemed correct. He wrote, "Yes that's certainly enough to say that there are significant signs of AI generation. it's long and high confidence."
He also shared with me that AI usage among thought leaders in op-eds is becoming increasingly common, and cited a study he co-authored demonstrating the growing trend.
One of the most common idiosyncrasies of AI generated writing is "it's not x, but y" framing. This appears repeatedly throughout the article:
- "The issue is not whether Palestinian suffering deserves recognition. It does. The displacement and trauma experienced by Palestinians during the 1948 Arab-Israeli war form an important chapter of Middle Eastern history deserving serious examination."
- "The catastrophe was not simply that Palestinians became refugees; it was that the Arab campaign to extinguish Israel had failed."
- "That is not a neutral historical observation. It is a false political proposition."
- "The refugee crisis of 1948 did not arise in a vacuum, nor was it simply the inevitable consequence of Israel’s creation. It followed the rejection by Palestinian and Arab leadership of the United Nations partition plan, which proposed both a Jewish state and an Arab state."
- "That is not education. It is curation by omission."
- "Museums are not activist or propagandistic platforms. They are custodians of public trust. Their role is not to inflame but to illuminate; not to advance ideological narratives but to encourage inquiry, historical nuance and civic understanding."
- "A museum devoted to human rights need not avoid difficult subjects. But it must present them with evidence-based inquiry, context, intellectual honesty and moral seriousness."
I believe this information merits an investigation into whether or not the article should be flagged as being AI generated. Even if it had a human hand in its crafting, it should be appropriately labelled and explained.
Without an investigation, Globe and Mail's readers will be forced to question how seriously the newspaper takes its new AI policy. I have never come across an article in the Globe and Mail that so heavily featured the signs of AI generated writing, and do not bring this forward lightly.
Please let me know how you will proceed.
Best regards,
Jason Toney
Do you like this post?
