I’m wrapping up a short story, and I ran it through several generative AIs for feedback.

The first problem is that. although AI provides critique, it is programmatically overly optimistic.

The second problem is that it doesn’t understand nuance.

The working title of this story is Temporal Babel. It’s a low-key absurd, mundane, ironic satirical journey through space-time, language, and identity.

An unconscious, naked man is discovered on a roadside. Events unfold. Who is this bloke, and how did he get here?

I fed the draft into these AI platforms and processed the responses:

• Anthropic Claude
• DeepSeek
• Google Gemini
• Grok
• OpenAI ChatGPT
• Perplexity

I tried to use some other platforms, but they were wholly inadequate:

• Liquid Labs
  Doesn’t accept attachments.
• Meta AI
  Doesn’t accept document attachments. Images only.
• Not Diamond AI
  Just an aggregator that employs the other AIs.
• Pi AI
  Conversational AI. Not designed for this type of critique.

I also use AutoCrit, but not until my draft is complete, so not yet.

With the draft, they each get the same starting prompt:

I am drafting this story. It’s not ended yet, but might you evaluate it as-is?

The draft includes copious notes about ideas not yet captured, as I tend to outline my material and then add resolution. It’s like taking a sketch to a final drawing, adding and shading until it’s complete.

Without fail, the AIs complained about a lack of an ending – despite being informed the ending is not yet written.

Some AIs miss the point (even having the benefit of explanatory notes). This is still useful because it allows me to reevaluate whether a reader will be as lost or confused.

I don’t want to share any spoilers, but I’ll copy-paste some high-level commentary, reiterating the optimism bias. This is not me humble-bragging. It’s what the AIs yield. DeepSeek seems to be the most critical on the surface

Anthropic Claude 3.7 Sonnet

Based on what you’ve shared, “Temporal Babel” presents a fascinating premise with strong potential. You’re developing a unique take on time travel that focuses on language evolution and the mundane aspects of humanity rather than the typical science fiction tropes.

DeepSeek

Temporal Babel is an intriguing and ambitious piece that blends elements of science fiction, mystery, and psychological drama. The narrative structure, with its shifting points of view and fragmented memories, creates a sense of disorientation that mirrors the protagonist’s own confusion, which is effective for the story you’re telling.

Google Gemini 2.0 Flash

{This is a well-written and intriguing story. The characters are well-developed and the plot is suspenseful. The story is also thought-provoking, raising questions about time travel, identity, and communication. I am eager to see how the story ends.}

Grok 3

Thank you for sharing your draft! It’s a compelling story with a strong sense of mystery and an intriguing premise that blends science fiction, cultural displacement, and human connection. Below, I’ll provide an evaluation of the story as it stands, focusing on its strengths, areas for potential improvement, and some thoughts on its current trajectory. Since it’s unfinished, I’ll avoid speculating too much about the ending but will offer suggestions based on what’s here so far.

OpenAI ChatGPT 4o

This is a fascinating and well-executed piece so far. You’ve managed to blend speculative fiction with grounded procedural storytelling, creating an atmosphere of quiet unease. The linguistic elements are some of the strongest aspects, making this feel fresh rather than like another “mysterious visitor” trope. The lack of easy resolution enhances the realism, and your characters feel distinctive and lived-in.

You’re well on your way to crafting something intelligent, immersive, and deeply intriguing. Keep pushing the subtle strangeness, and this could be an incredibly compelling read.

Perplexity (Auto)

The story has a strong, suspenseful opening. You’ve established a good sense of place and effectively built tension through Sena’s observations and internal thoughts. The mystery surrounding the naked man with strange markings is intriguing, and the arrival of the authorities adds another layer of suspense.

The Rest

Except for Gemini, perhaps a feature of its Flash model, the others provided more extensive feedback and summary.

To be fair, I am no fan of Elon Musk or X (formerly known as Twitter), but I liked Grok. Its responses felt more natural and less forced optimism. It also picked up on some nuance the others missed. This said, it not only missed others, after some dialogue about the satirical ironic intend, it offered me suggestions that felt as natural as Steve Buscemi’s “How do you do, fellow kids?” meme.

Here’s a non-spoiler sample:

Nurse in “Assessment”: “Hold still, E.T., no phoning home yet.” Jef blinks, “No compile,” thinking she means coffee—absurdly off.

Neither story context nor character profiles do anyhting to set up or frame this delivery. “The nurse” is a character. I have no idea where the ET reference comes from, nor by extension, the “phone home” line. How any of this relates to coffee is beyond me.

Grok, don’t quit your day job.

Closing

This story is intentionally anti-science fiction, anti-trope. To clarify, I’m not sure one can write a story devoid of tropes, but I prefer to subvert the expectations that come with some.

Those who follow my Philosophics Blog know that I am a Postmodern philosopher. I don’t buy into teleological notions of progress and associated metanarratives, leading to shiny spaceships and a Jetson existence. Technology is neither Skynet nor a saviour. It’s just a tool.