How I Got Identical Answers from ChatGPT and Gemini

It’s proof that you can get the AI to be super accurate leaving no room for hallucination.

Image Created by Author using Midjourney

Few days ago I needed to create a mini guide on a scientific topic, which I initially researched. It was needed for my 9–5. 

I decided to experiment using both AIs just to see how they perform. I used the following models:

• ChatGPT 4o

• Gemini Advanced 

I have paid subscriptions for both. 

I started a conversation with both AIs, using the same prompts.

What I did:

1. I searched for trustworthy literature. Gathered 10 high-quality sources — research, articles, experts opinions, case studies, company reports. Downloaded them all to a folder. 

2. I had an outline in mind, so I drafted it. 

3. I gave initial instructions to both AIs about what this chat was going to be about. Asked for confirmation it understands. 

4. Then asked to analyze the literature, take my outline, and create a table of contents based ONLY on the information I provide. With this prompt, I uploaded the 10 files I downloaded and pasted my outline in the prompt. 

5. To my surprise, with a few minor differences in the exact wording, the answers from both AIs were identical!

I continued with the chapters. The process went on and the answers stayed consistent and close to one another up until the conclusion section. Sometimes I had to choose between two different wordings of the same thing. 

The two AIs can nicely complement each other in such tasks. 

But why is that important: to me, it’s proof that you can get the AI to be super accurate leaving zero space for hallucination. 

How can this be applied to writing online? I have the system. I’m turning it into a digestible guide for someone who never touched those two AI guys. Stay tuned. 

Thanks for reading!

Image Created by Author using Midjourney

👉 DOWNLOAD my FREE Writing Tools I use to make consistent money on Medium.

[This story was initially published on Medium.]