👤 Authored by Lashmi Bai Ravindrapandian
📅 Published on Mar 7, 2026
I have always found those celebrity “Ask Me Anything” sessions interesting. Someone comes on YouTube or Instagram and says, “Ask me anything.” People send questions, the celebrity picks a few and answers them.
I thought… why should celebrities have all the fun? So I started doing my own AMA sessions with AI. But there is one small difference…
I am the celebrity. AI is the one asking me the questions.
This actually came from my experience with AI hallucination. AI is very good at giving answers… sometimes a little too good.
If my prompt is not clear, AI may make an assumption and continue. If some information is missing, it may try to fill the gap.
And the answer can still sound very confident. That is where the problem starts. I don’t want AI to quietly decide what I meant. I want it to ask me.
So now, when I give AI an important prompt, I ask it to first review my request and run an “Ask Me Anything” check.
I want it to look for anything that may cause confusion later. Is something missing? Am I making an assumption? Is something unclear? Are two instructions conflicting? Is there a risk I have not thought about? Is there an important decision I have not mentioned? Are there any edge cases? Could anything in my prompt make AI guess or hallucinate? If yes… ask me first.
Readymade prompt I use
This is the instruction I add:
At any point in the conversation or during a context-heavy session, if you feel an AMA-style clarification is needed, pause and ask me questions before continuing.
Review what we have discussed so far and check whether:
- something important is missing
- you are making an assumption
- the requirement is unclear
- there are conflicting instructions
- a decision has not been made
- there is an edge case I may have missed
- the context has become too large or confusing
- there is any risk of guessing or hallucinating
If needed, start a quick “Ask Me Anything” round and ask me the questions you need. Do not silently fill the gaps yourself.
Once I answer, continue with the task using the updated context.
If everything is clear, simply say “No clarification needed” and continue.
The AMA is not limited to the beginning of the task. You can trigger it at any point whenever clarification would help you do the work correctly.
Does this actually work
Earlier, I would give a prompt and expect AI to figure out the rest. Now I am trying to change that habit. If I have not given enough information… I want AI to tell me. If I have left something open to interpretation… I want AI to ask. If it sees a risk… I want to know before the work starts.
And sometimes AI asks a question and I realise… “Oh… I didn’t think about that.” That’s useful.
This approach can help reduce one common problem… AI making assumptions when I have not given enough information.
The question itself becomes part of my learning.
My takeaways
For me, this is not about creating another perfect prompt. It is about not making AI guess when I can simply give it the right information.
And sometimes…
the best answer starts with a question.
Discover more from LR Virtual Classroom
Subscribe to get the latest posts sent to your email.