“Should lecturers teach students how to prompt with Gen AI?”
This was a question raised during our panel session at the AI Summit, organised by the Centre for Information Integrity and the Internet (IN-cube), NTU.
Prompting is basically telling the machine what you want. If you know what you want (intent), and you can evaluate the machine’s output (expertise), then all is good. The problem is that if you don’t really know what you want and if you lack expertise, then no amount of prompt engineering training can help you get your ideal outcome.
So I replied, “There are two things that we can teach – domain expertise and prompting. The question is which comes first?”
It seems obvious that we should teach domain expertise first, but after thinking through the question after the symposium, I think “prompting” needs to be tackled differently.
Prompting is writing and if you’ve been following my content, I always say that writing is thinking. So if we are to “teach prompting”, we need to actually teach people how to think better.
Prompting <-> Writing <-> Thinking.
But what should we think about? First, our domain of expertise. There’s plenty to think about within a domain (say geography), but the best thinkers think across domains (mixing geography and the arts). Thus, you need to broaden your skills and knowledge, and learn frameworks like design thinking and systems thinking, and then…
I shall stop here. There is much to say on this matter.
For today, what I’m trying to say is that teaching AI literacy is harder than it appears. Unlike Microsoft Word, where you can learn how to use every function via a user manual or structured lessons, Gen AI has no fixed user manual – it really depends on how you can think, judge, and articulate your thoughts.
What I wrote this week
I fell into another AI rabbit hole and two hours of my day instantly disappeared.
Do read Ethan Mollick’s piece on “Management as AI superpower”.
I bet you didn’t know the origins of the phrase “OK”. I didn’t!
A small wish list for Microsoft that may avert the AI bubble burst.
Me trying to game the LinkedIn algo with dense infographics, after observing others who are doing so.