AI-Powered Data Analysis: A Practical Introduction Q&A with Tina Lasisi

AI-Powered Data Analysis: A Practical Introduction from Michigan Online is designed to open doors to data analysis, whether you're a beginner asking your first question or an experienced analyst exploring new AI tools. This interview with course instructor Tina Lasisi highlights her inspiration for creating the course and how asking a good question can be empowering. 

AI-Powered Data Analysis is your first course with Michigan Online! What sparked the idea of creating a course about data analysis and AI?

I wanted to create a data analysis course on AI because one of the biggest use cases of generative AI is in the coding aspect of data analysis. One of the biggest hurdles people encounter when conducting data analysis isn’t usually what they want to accomplish; [it is] figuring out all the quirks and specific features of the [programming] language. That takes some time. With AI as a copilot, we don’t have to worry about that as much. 

How can generative AI be a tool for those just starting in data analysis? 

The first thing that we go over is what exactly is data. How do you get data? How do you interpret data, and what kinds of statistical questions can you ask? It's a very high-level view, just understanding how to formulate a question. A lot of people could get lost in the sauce if they don't know what the actual question is they’re setting out to answer. There are many ways to get there, and you don’t have to be a mathematician, a statistician, or a computer scientist to ask good questions. 

How can it help those who are already more advanced in data analysis? 

For those who are more advanced, it’s just showing them that using AI can be intuitive. I genuinely believe that there are many things that are presented as harder than they need to be. It's really important to close that rift between people outside of data analysis and those in it, because you end up speaking a completely different language. So a lot of this is me approaching it as an anthropologist, thinking, what are these different cultures? What are the different histories and vocabulary that people use in these different groups? And how can we bridge them?

In the course, you discuss how generative AI can be a powerful tool for conversations, information retrieval, and generating solutions. Why do you think these are important when approaching a hypothesis or problem? 

How many of us have gotten comments back on assignments we’ve submitted, where we’ve been told, “Great, you have the answer, but you need to show your reasoning.” A conversation shows you the value of being able to articulate what you want to do and how to do it for yourself. As for information retrieval, there have been a lot of times when I vaguely know where something is, but being able to retrieve it is quite important. With AI, you can now do much more complex information retrieval, so figuring out how to use it in that capacity is crucial. Generating solutions [with the help of] AI is about breaking through that blank page paralysis. Thinking critically is essential because you can end up in a loop with AI. 

All these things come up when we ask questions and problem-solve, so thinking about them while working with AI can help us get better answers. 

You previously did these really entertaining science explainer videos with PBS on how our bodies work, and you brought that energy to this course as well! How did you cultivate your approach to explaining more complex concepts for learners to better connect with and understand the material?

I originally did not come from a science background. I grew up in the Netherlands; you had to decide what area you wanted to specialize in at 14. I went in the humanities direction because when I looked at the math and science tracks, I thought, “Oh, that seems difficult.” I ended up in cultural anthropology. But along the way, I ended up asking a question, and I was supported, empowered to ask: “Why did humans evolve scalp hair?” You don't need to know calculus to ask that question. 

And yet, I have been surprised to find in the last 10 [to]15 years people who do not have a science background will approach me and ask those same, really interesting questions and say, “Oh, I'm so sorry. This must be a stupid question.” I, a person who was outside of science, asked a question just to know, and found that there were no answers to it. That set me on this path. These aren’t stupid questions. 

I want to give that power to other people, because there's a lot of value in curiosity. I think a lot about how unfortunate it is that a lot of education, and not just STEM, but education overall, is structured in this way that is more about weeding out than bringing people in. How sad would it be to think that we're missing out on incredible scientific insights because we're selecting for very specific people who ask questions in really specific ways to be the only people who are deemed worthy enough to contribute to science. 

How do you think using generative AI for data analysis can help people advance in their careers? 

I think that in the near future, not using generative AI for data analysis will feel like doing statistics by hand with those old critical value tables, where you had to look up significance levels. I’m not that old, but I do remember one of my stats professors insisting we learn to do it by hand so we’d “understand what was going on.” I can’t say that’s ever felt helpful to me, or that I’ve used it since.

And when we say “using generative AI for data analysis,” that already includes a lot. Data analysis isn’t a single, coherent task. It involves importing your data, formatting it, writing code, commenting that code, and troubleshooting when things go wrong. Generative AI gives people access to more advanced tools without requiring them to master every technical detail upfront.

What is the most important lesson you’d want someone to take away from this course? 

The most important thing is that if you’re coming in with no data analysis background, I want you to get better at asking questions. If you’re more advanced, in terms of the diversity of the tools, I want you to know you can troubleshoot and experiment, knowing that a lot of times, the thing that’s stopping you is the software, not you. 

Prior to designing this course, I had never used Jupyter, but I promise you that my entire experience with Jupiter Labs has been powered by AI. I want to encourage people to approach this technology as a safe playground, a sandbox, to try things out, and know that there is really no harm in trying. If you manage to navigate to the Michigan Online website, you can learn how to do a lot of data analysis! It’s all about being exposed and becoming familiar.

AI-Powered Data Analysis: A Practical Introduction

3 weeks

Course

As generative artificial intelligence (AI) reshapes our world, the ability to analyze data is quickly becoming as fundamental as reading and writing. “AI-Powered Data Analysis:…