How do I begin querying generative AI?
As the early adopters of generative AI in our member organizations are working with their colleagues to develop generative AI skills, one of the hardest adjustments for newbies is remembering
It's not Google.
It's not for searching. It's for helping you do a task.
We have to unlearn how we ask questions of Google, in order to learn how to work best with an AI assistant.
One of the easiest ways to make this mindset shift is to try prompting generative AI with personal tasks. Things you'd normally ask Google, like
- What are some examples of a great best man's toast?
- What are the best places to stop on my road trip?
- How do I have this difficult conversation I've been avoiding?
With Google, you ask once, get a bunch of results, browse all the results, and then scan each of them to hunt for an answer to your question.
With a generative AI assistant, you get a single result, and then you keep iterating until you get the answers you needed.
To get you started, here are a few examples from my own prompt history. (Note that I'm using ChatGPT's premium version, but any AI assistant will work similarly)
Click on a link to see the back-and-forth conversation in ChatGPT.
Scenario
|
What I learned
|
I can ask my assistant to try again, in lots of different ways, with instantaneous results. This prompt took me about 10 minutes.
|
|
My friend called with a challenging problem at her work. I didn't remember enough about Crucial Conversations in order to help her. I gave my assistant the scenario and got an outstanding application of the Crucial Conversations model. Published models + my context = impressive assistance.
|
|
I wanted to suggest a "Secret Santa" activity but make it more globally inclusive. None of the responses alone was quite what I wanted, but I got enough to combine a few replies into my own table. And I learned a lot!
|