Post

Saw a data analyst ad that had "Familiarity with ChatGPT or other AI tools for data analysis" as a nice to have 🤪 I guess we're at the point where employers think ChatGPT is a tool for data analysis. Cool cool cool cool cool. #databs #datasky

It is not?

Haha sorry, it is bait but for an answer I'm genuinely interested in. AI tools (including Chatgpt) offer data analysis features at this moment, I'd consider some of them (with huge caveats) to be helpful. And I'm curious if your post is a humorous take on too much AI in DS or of AI in principle.

My problem is the distinction. AI tools absolutely can help with data analysis (e.g. with SQL or Python syntax or whatever), but they are not tools *for* data analysis. They cannot analyse data for you because they cannot think about problems properly. Though I do intend to test this more this year.

Yeah, I am kind of interested in this. I could see you being mistaken about their capabilities though, depending on your definition of analysis. Haven't tested this much, but they would probably be able to do exploratory work, help a lot with understanding data and can also answer clear hypotheses.

For a concrete example I did last year (?), I ran an open data set from a portal through ChatGTP to explain the schema. It did so, while sourcing the official documentation from the German railway operator (that was part of the data set release but got lost when being federated to the portal).

So in that case it was helpful (though you might disagree that that was a step in data analysis) and provably correct (because I could find the official docs). I think the larger issue is if you have no way to verify what an AI system tells you, then you run into hallucinations and mistakes.

In any case, sorry for the bait! But I thought it's an interesting question 😅.

I was joking with the bait thing 😁 what you describe is the tool *helping* with data analysis, not *doing* data analysis. It can answer clear hypotheses, but I'm not convinced it can generate them. So it's an aid for someone who already knows what they're doing, not an analyst replacement.

So if the original job ad meant they want someone who "uses AI tools to speed up the time consuming parts of the job" then I agree, that's a good skill to have. But if they meant "gives the entire analysis to ChatGPT to get answers faster" then that's a huge red flag.