Em Dash Man

Over the past year, I’ve found myself observing and participating in the online discourse around generative AI. It’s a fascinating mix of optimism, caution, and—often—overcorrection. One recent pattern I’ve noticed among the more anti-AI voices is a tendency to declare that certain quirks of writing automatically “reveal” that a piece was written by AI.

Top of the list? The em dash.

​Apparently, if you use em dashes in your writing, there’s a decent chance someone online will accuse you of being a bot—or worse, of using ChatGPT.

​Earlier this week I read ‘Revenge of the Librarians’ by Tom Gauld, a fun and nerdy book that includes a comic about ‘Apostrophe Man’. Apostrophe Man is a literary superhero whose superpower is punctuation. He is both absurd and strangely relatable (maybe not so strange to those of us who like (or feel the need) to correct grammar). The combination of that comic and the em dash paranoia online sparked an idea: What if there was a superhero like Apostrophe Man, but for the em dash instead?

​Thus, Em Dash Man was born.

---

​I played around with ChatGPT to create both images and comics based on this idea. Below are two images of Em Dash Man. One in a retro comic book style and another in a dramatic fantasy realism style. 

Retro Comic Book Style

Dramatic Fantasy Realism Style

---

​I kept playing around and iterating, eventually cooking up the two comic panels below. I'm not claiming they’re masterpieces, but the process was fun and enjoyable. I always try to take the opportunity to experiment with ChatGPT. It’s a great way to learn, and I encourage others to do the same. 

​There’s something cathartic about channeling these online arguments into something silly and symbolic. Sometimes satire says more than a thread ever could. 

---

EM DASH MAN
​—DEFENDER OF THE WRITTEN WORD!

A tale of accusation, punctuation, and redemption.

---

EM DASH MAN AND THE GREAT AI DEBATE

AI panic hits the town hall. Only one hero dares to intervene. 

---

A Note on “AI Tells”

​It's true that generative AI models can show stylistic patterns in their outputs (em dashes, choice of quotation marks, a certain rhythm of prose, even emoji use). However, those “tells” are not universal and are often the result of vague prompting.

Generative AI is highly responsive to direction. You can ask an AI model to:

• Write without em dashes.
• Use only commas or short sentences.
• Format answers as bullet points.
• Use MLA or APA formatting.
• Mimic any tone, author, or style.
• Format like a technical manual or a short story.
• Write like Carl Sagan, a 2007 MySpace page, a casual Reddit post, or a 1950s advertisement. 

​In other words: AI output is responsive and deterministic. If it’s outputting dashes, it’s likely because the prompt (or the training data it learned from) included them. 

---

Final Thought

​Generative AI is a deterministic tool based on math, but different prompts and additional training can change responses. This is similar to how human language evolves over time through usage, exposure, and cultural shifts. The more people engage with different writing styles, the more they absorb, adapt to, and emulate those patterns in their own expression—whether through comments on social media, reading chapter books, or now, through exposure to generative AI outputs.

​Em Dash Man isn’t here to save AI or destroy it.

He’s just here to say:
          “Maybe we should judge ideas, not dashes.”

---

All images in this blog post were generated by ChatGPT through my own prompting and iterative refinement—just one more way to explore what these tools can (and can’t) do.