Not Every Problem Needs an AI Solution

Jan 17, 2025

Hello you :)

Today’s letter is inspired by my just concluded Google Cloud PMLE exam.

This is how I looked after the exam (happy that I passed but my brain was all “Okay, can I get some rest now?”):

Something snapped in my brain when this happened.

I was planning to take my Google Cloud PMLE exam and needed to quickly figure out how many days I had left before my voucher expires.

So I turned to my friend and colleague and said, “Hey, can you help me calculate 31 minus 7?”

“Sure,” he responded casually, pulling out his phone.

I watched, expecting a quick reply. Five seconds. Ten seconds.

The answer had already come to me in fact, but I didn’t want to be rude and just say, “Never mind, got it” so I kept watching and waiting.

A whole minute passed, and still — nothing.

Finally, I couldn’t hold back.

“What’s taking so long?” I asked, genuinely puzzled.

And then, he said it — The words that completely floored me — “The network is poor. ChatGPT is taking forever to load.”

I was short for words. ChatGPT? For basic subtraction?

Something so simple that even the most ancient, dust-covered calculator could handle without blinking?

“Oh, I see. Don’t worry about it then.” I finally found some words to say and turned away.

AI/ML is a great solution to many problems today. But is it always the best solution?

Look at the case of missing values in data. Is using ML algorithms for imputation always the best way to deal with the missing data?

The short answer is no.

As a big AI/ML fiend, it’s hard to admit it, but ML is not always the solution our problem needs, and that’s okay.

Spend at least 10 minutes every day learning something data-related that you didn’t know before.

GDG Babcock Chatbothon: A hackathon for chatbot solutions only sponsored by Ready Tensor.

DATAPRINCESS TO YOU