I started off by asking Gemini AI to give me questions that would help me experiment with AI in research. link to the chat

I  chose this question

Question: “How does a 4-day work week affect employee productivity in the tech industry specifically? Find one study that supports it and one study that highlights the risks or downsides.” Why it’s good: By asking for “one pro and one con,” you are testing the AI’s ability to be neutral and find conflicting data.

I started by using a research rabbit and asking it what it is good at, it turns out it is not like Gemini, ChatGPT, or Perplexity. It gave me research docs only.

I began experimenting with research rabbit with this question, and the results were that he gave me more general results on the effect of working 4 days rather than on the tech industry. Some of the results were also not from credible sources, and I could not figure out how to tell him to categorize them only to credible sources. I think the key strength is that it makes a map showing connections between the research docs. Here’s the link to the chat

I then started to explore the question on perplexity (my favorite AI tool, probably the only one I use). It categorized the question into 3 sections. It gave me relevant answers to the questions and only focused on the tech industry. Here is the link to the chat

I asked Gemini to do a comparison table based on what I wrote

I think that if someone will rely 100% on AI in doing research, research rabbit alone is not enough. It could be used for knowing the sources only and then these sources can be added to perplexity for followup questions.