Using ChatGPT as a Data Science Manager

From the yellow, big, fat phone book to telling Siri to make a call for me..

The Past

The world has transformed a lot since my early days as a data science individual contributor. I recall spending hours digging into the initial patterns utilizing dplyr and ggplots, building models based on statistics textbooks, creating custom R packages to address specific questions, and googling problems to find solutions from the community in stackoverflow.

The New Norm

Today, we see data scientists kick off their projects faster by utilizing ChatGPT in various ways. Before I go into this, let me share a time when it struck me the new norm has changed.

One time my team was having a social lunch in the office. We were talking about different pets, and some historical backgrounds of certain dog breeds. When a question arose, the “new folks” quickly opened ChatGPT for an answer, while some of the “old folks” were still Googling and refining their search terms.

It’s fascinating to see how productivity has increased with data scientists as they can quickly search for references on how to solve a problem, jump start on coding by asking ChatGPT for generic code for Python, and solve coding blockers through AI built-in tools like Databricks. Today, you can even refine your story telling skills, which is a key aspect of data scientists, through ChatGPT. (and of course, these are all administered within the company’s AI policies)

So How am I adapting?

As a manager of managers and leader of multiple teams, my role has shifted from building models and conducting analyses to summarizing team learnings and developing strategic narratives along with other functional leads. Here’s how I have been leveraging AI tools and ChatGPT to enhance my effectiveness:

• Summarizing key points across multiple analyses: I use AI tools to syntehsize multiple analyses from various data scientists. This has helped me identify themes and sources quickly to inform the next strategy.
• Identifying trends in raw data: By inputting raw data (within leagal and privacy guidelines of the company), I am able to get high-level patterns fairly quickly using AI / ChatGPT’s EDA and modeling capabilities. This has accelerated me to point to my team which areas to dig in further to develop better narratives.
• Crafting strategic narratives: As a leader who frequently communicates with the larger groups, AI has helped to adjust my message tone to ensure clarity and engagement.

The future

I don’t believe ChatGPT will replace data scientists’ job, but rather enhance the productivity and accelerate decision making wherever applicable. In ejoy using it occasionally to kickstart new projects, even if I don’t fully rely on its accuracy (that’s when data scientists come in to go deep). Embracing these tools excites me for the future of data science.