Does Gemini Generate Humanized Writing?

Summary

The author discusses their experience with testing Gemini's writing generation capabilities, finding that it produces human-like writing that often passes as human-generated when assessed by GPTZero, contrasting it with ChatGPT which requires more training and editing to achieve similar results.

Abstract

The author of the web content has conducted several tests on Gemini, an AI writing tool, using GPTZero to detect AI-generated content. Surprisingly, Gemini's output consistently received a "human" assessment, even with limited context or training. This outcome stands in stark contrast to the author's experience with ChatGPT 4, which needed extensive training and editing to produce content that could pass as human-generated. The author has documented these findings in a side-by-side comparison, highlighting the quality of Gemini's writing and its implications for AI-generated content detection tools like GPTZero. The author speculates on the possibility of Google having created an advanced AI capable of humanized writing due to its vast data resources, while also considering the alternative that GPTZero might not be fully adapted to recognize Gemini's new model. The article concludes with an invitation for readers to follow the author's journey on Medium and a call to action to download a free guide on making money online with AI for creators.

Opinions

The author was initially skeptical of Gemini's ability to generate human-like writing, considering it might be a glitch when GPTZero assessed it as human writing.
Repeated tests reinforced the initial findings, impressing the author with Gemini's consistent "human" assessments.
The author expresses that ChatGPT requires significant effort to produce content that mimics human writing, unlike Gemini.
There is a hint of concern about the ethical implications or potential "scary" aspects of an AI that can generate humanized writing, especially one backed by Google's data.
The author questions whether the unexpected results might be due to GPTZero's limitations in adapting to new AI models, rather than Gemini's superiority.
The author is optimistic about the potential of AI in content creation and encourages readers to engage with their ongoing exploration of AI tools and strategies.

I tested it for AI-generated content and I was shocked!

Not a member? Read this story here.

Image generated with Midjourney

A few weeks ago, I tested Gemini with GPTZero. I got a pretty confident assessment for human writing.

I got confused…usually, GPTZero catches the pure AI generation, so I thought it was probably a glitch.

But then I thought “What if…”

And I tested again…and again…

And…

Screenshot by Author

This can’t be true, I thought…

So recently I tested ChatGPT 4 and Gemini Advanced and created a side-by-side comparison.

I wanted to see the difference between the models, so I asked for a super short Medium article.

The prompt and the result from one of the tests. Screenshot by Author

This was just one of the use cases, but I was impressed by the quality of the writing without a single training or enough context! With ChatGPT I had to do extensive training to achieve something close to human and it always needs editing.

But for Gemini, all of the tests produced HUMAN assessment with GPTZero! True, some with less certainty and less than 100% but still HUMAN! Read about my test here:

The thing is: everythime I test Gemini’s content, it’s HUMAN!

I tested it for AI-generated content and I was shocked!

This can’t be true, I thought…

I Tested ChatGPT4 and Gemini Advanced with 3 Identical Prompts

Medium short-form article, X (Twitter), and LinkedIn post. I checked for AI-generated content and readability too! The…

Is this real?

What do you think?

Thanks for reading!

Yana

The One Reason Why I Choose Gemini Over ChatGPT

And You Should Too!

Read or Die — Publication Rules

Updated January 2024 Guidelines