How to use AI properly in research

• Four Tips for AI Usage in Research
  • 1. Express yourself without AI
  • 2. Always write your first draft yourself
  • 3. Be careful with AI-as-a-judge
  • 4. Practice juding AI’s response

Four Tips for AI Usage in Research

1. Express yourself without AI

Many people think AI can help them polish their language, but this is not always true. AI involvement can dilute your own voice and contribution. Instead of making your ideas clearer, it can act as a layer of mediation that blurs your original intent. As a result, what you ultimately present may not fully reflect what you actually mean.

2. Always write your first draft yourself

Writing down helps clear your mind. This makes sure you understand what you are working on and what you are writing, so you can deliver your work to others clearly. If you write something and someone asked you a follow-up question, you need to ensure you know the answer and do not waste others time. You should not pretend to know, and AI is not responsible for your words.

3. Be careful with AI-as-a-judge

AI is so good nowadays that many people use them in data analysis, such as annotation. However, the number should have some meaning. AI can just give you any number, and it can give different numbers if you try many times. So it is important to have a well-defined and concrete rubric to follow. This makes your number more convincing instead of randomness.

4. Practice juding AI’s response

AI is good at confusing you. It may generate a long long response, which seems so logical that convinces you. However, you should understand your work, or at least some parts of your work, so that you can tell whether AI is correct or it is just making meaningless suggestions.

I found this Hewitt’s blog really worth reading. Good luck with your research!