Opinion: limit ChatGPT to three options and stop at 80% to keep work fast

I opened ChatGPT to start working, but

I realized 20 minutes had passed.

Has that ever happened to you?

"Make it a bit easier to understand."

"Give me another idea."

"Make it shorter."

"Actually, go back to the first idea."

Before I knew it, the conversation with ChatGPT just kept getting longer.

It happens to me, too.

I'm using AI to make my work easier, yet I'm spending all my time talking to the AI.

That's a bit counterproductive, isn't it?

Recently, I've started to think that deciding where to use AI is more important than just "using AI a lot."

ChatGPT will answer as much as you ask.

This is both its convenient side and its scary side.

For example, when thinking of a title for a document:

"Think of 10 titles."

10 appear.

"10 more."

More appear.

"Something with more impact."

More appear again.

Before I know it, there are 30 options.

But in reality, I only need one.

Increasing the number of choices doesn't necessarily make the work better.

In fact, having too many choices can sometimes increase the work of having to choose.

Recently, I've been limiting it to "up to 3 options."

So, what I'm conscious of is

narrowing down the options I ask the AI for from the start.

For example,

instead of

"Think of a title,"

I ask:

[For Copy-Paste]

"Please provide 3 title ideas for this document. Please vary the direction for each. No additional ideas are needed."

I can compare 3 options.

If there isn't a good one among them, I'll fix it a little myself.

It's like creating a standard for when to end the conversation with the AI.

"Once it reaches 80%, I'll fix it myself."

There is one more thing I'm conscious of.

Not demanding 100% from ChatGPT.

For example, an email draft.

The first response is 70%.

I fix it a little to get to 80%.

From here,

"Make it a bit softer,"

"Change this expression,"

"Adjust the ending,"

it's sometimes faster to just finish it myself than to keep asking the AI over and over.

Don't try to make the AI produce 100%.

Get it to 80% with the AI, and finish the remaining 20% yourself.

Since adopting this mindset, it's become easier to use.

Deciding "just one more time" is also recommended.

To avoid repeating revisions endlessly when the response is mediocre,

[For Copy-Paste]

"Please improve this response one last time. Based on the content so far, please narrow it down to the single most practical option and provide the final version."

I sometimes ask this.

The point is to decide for yourself that it's "the last time."

AI doesn't have a quitting time.

If you keep asking, it will keep answering forever.

So, you are the one who has to end it.

The purpose of using AI is not "to use AI."

When you start using ChatGPT,

you start thinking,

"Maybe there's a better prompt,"

"Maybe I can leave more to the AI."

But what I originally want to do is

not to use ChatGPT skillfully, but to finish my work easily.

That's what I believe.

That's why recently,

up to 3 options.

Stop at 80%.

Only one more revision.

I am conscious of these three things.

I want you to try this when using ChatGPT today.

If you use ChatGPT for work today,

when you think, "Maybe a better answer will come out,"

please think about it just once.

"Isn't it faster to fix this myself?"

If it's faster to leave it to the AI, leave it to the AI.

If it's faster to do it yourself, do it yourself.

That level of distance is fine.

If you want to make work easier with AI,

I want to shorten the time spent using the AI itself as well.

That's what I've been thinking about lately.

*When using generative AI for work, please check your company's rules and avoid inputting information that is not permitted, such as confidential information, personal information, or customer information.