Using Gen AI as an Idea Generator

One great use of Gen AI is to enhance your expertise by generating lots of ideas for you to consider.

Imagine, for example, that you’re writing an article titled “Using Gen AI as an Idea Generator” and you want to start with an illustrative example of how you can use Gen AI to generate ideas for some task. What example would you pick? This kind of thing used to be difficult in the old pre-GenAI days but has become much easier now. I just asked ChatGPT:

As you can see, it has given quite a lot of ideas. Not all of them are great, but there are definitely more interesting ideas here than I would have thought of on my own.

And this basic pattern repeats on an almost daily basis for me. The thing I’ve learnt is to make it generate an insanely large number of ideas, just increasing catchment area—sometimes far beyond what’s reasonable. If I had an assistant helping me, I would have asked for 3 or 4 possibilities. But for the AI I asked for 20. In some cases, I’ve gone up to 50 before getting what I wanted. And, as I keep saying, you should push ChatGPT more—be a demanding boss. “Give me more out-of-the-box ideas” “Give me ideas that would appeal to an Indian audience” “Give me 20 ideas all related to elections” and so on. You’ll be surprised at how many ideas it can come up with.

Most of those ideas will be crap but that’s OK. Generating the ideas takes just seconds and rejecting the dumb ones also just takes seconds. Even if 2 out of 50 ideas are good, that’s 2 good ideas.

And don’t take my word for it, there’s now research to prove this point.

A study of 1,018 scientists in the R&D lab of a large U.S. firm showed that AI-assisted researchers “discover 44% more materials, resulting in a 39% increase in patent filings and a 17% rise in downstream product innovation” compared to researchers not using AI.

More interestingly, the impact of AI isn’t the same for everyone: “while the bottom third of scientists see little benefit, the output of top researchers nearly doubles.”

How does this work? AI generates lots of ideas: some good some bad. The smart scientists use their domain knowledge/instinct to quickly zero in on the good ideas, while the not-so-smart ones waste a lot of time exploring the dumb ideas.

Take-home messages:

• Use of Gen AI as an idea generator can give big productivity boosts

• You become an idea evaluator/curator/refiner instead of an idea generator

• Domain expertise continues to remain important in the age of AI