Should I Fire My Analyst?

I expected much better from him.

The job wasn't that challenging. If I had done the job myself, I would have saved some time. More importantly, I would have avoided the pain of explaining the same thing time and again with the impression of banging my head against a wall.

But I was feeling lazy.

And what's the point of hiring an analyst if you cannot delegate some tasks to him that you feel lazy about?

Anyway, before making the call, let me tell you the job I asked my analyst to complete:

1. Analyze a .csv file downloaded from Apple Podcasts showing the number of listens of a podcast.

2. From all listens, calculate the % of listeners in the main cities in Spain.

3. Show the results on a map of Spain, displaying a dot on each city whose size is proportional to the % of listeners.

Sounds simple enough, right?

Well, not for him.

Here is what he came up with.

I’d have expected a more spectacular performance from my analyst. Something that would make me proud and show off his abilities. But after the hundredth time trying to improve the outcome, I gave up.

Should I just accept his limitations and never assign him a similar task again?

Oh, come on!

The data in the .csv file was clean and organized. The file didn't consist of millions of rows, not even many thousands. There were no problems with dates or formats.

If he can’t handle such a basic task, what is he good for?

Especially knowing all the hype about his superpowers and super capabilities that are supposed to send us all to the unemployment office in the near future.

In case you’re wondering, my analyst's name is ChatGPT, and his surname is Advanced Data Analysis (formerly known as Code Interpreter).

I don't remember how many prompts I tried, how many apologies I got, and how many megabytes of crap I downloaded before getting a decent result.

Do I suck at prompt engineering, or rather does AI in general and ChatGPT in particular still have a long way to go until they can rival us humans?

But I don't want to be too harsh on him.

After all, he kind of delivered the result I needed.

As I said before, I was expecting something more spectacular that I could brag about on social media, but he got the job done.

And to be honest, it would have taken me much longer to achieve a similar result.

Instead of wrestling with Python code, or figuring out how to generate a similar graph with PowerBI, I just limited myself to giving ChatGPT some instructions in plain English.

And then, after some iterations of swearing and complaining, he delivered a result I could use.

You know what?

I’ll give my analyst another chance.

Looking back, I got the job done and also got the inspiration for this newsletter.

Thanks for the newsletter inspiration, ChatGPT!

By the way, the podcast I downloaded statistics from is “Sé feliz donde estés”, by Gache Boccazzi (in Spanish only).

The statistics show that this podcast is rocking it!

Thousands of people are already subscribed, and there are plenty of reasons for it.

The insighful yet humorous interviews that make you think and laugh. Her reflections on happiness and life that keep you engaged. Gache’s smooth voice with her charming Argentinian accent.

Just give her a try, and let me know what you think!

You can listen to her podcast here.

P.S. Don’t forget that if you’re in need of financial modelling services, you can always hit reply and let me know.

• If you liked this email, forward it to a friend.

• If this email was forwarded to you and want to receive similar ones, visit agilibus.