This experiment aligns with my recent experience using Google Gemini. In the past couple of weeks, I’ve shifted to using Gemini for most tasks, moving away from Claude (see how I use it) and ChatGPT.
I’ve noticed that Gemini doesn’t seem to grasp prompts that are written in a conversational style as well as ChatGPT and Claude do.
What I mean is, with Claude or ChatGPT, I don’t have to think about writing for a machine; I write as I speak, and it usually understands me.
However, Gemini sometimes gives responses like, “I can’t help you with that request,” or provides a technically correct answer that misses the conversational aspect, as you’ll see in the example below.
Let’s test the humor of Gemini, ChatGPT, and Claude:
Prompt: Tell me you are an AI without telling me you are an AI
ChatGPT Response
Not my kind of humor, but it gets the intention of my prompt. This answer considers obviously that I work in marketing and content.
Claude Response
This response shows the breadth of capabilities of AI assistants, kind of funny.
Gemini Response
Technically correct, but not what I was intending to get back from it. Gemini is not even trying to be funny or conversational.
Gemini’s key advantages, in my view, are its integration in GSuite workspace apps (Gmail, Docs, and so on), and its large context window. But when it comes to the overall experience, Google Gemini is lagging behind based on my experience of intense usage the last couple of weeks.
