Where Is Data Science Headed in 2023?
Think back to a year ago. Could you have predicted that by the end of 2022, tools like ChatGPT and Stable Diffusion would take the internet by storm, or that tech giants would lay off thousands of workers, including many data practitioners? Probably not.
No time-series forecast can tell us what the year ahead has in store for the constantly shifting fields of data science and machine learning. What we can do, though, is listen to experts who closely observe their respective communities and keep their fingers on the pulse of innovation. That’s precisely what we invite you to do this week: here are three excellent, future-leaning articles that make educated predictions for 2023, each in a specific domain.
- How will we do data science this year? If you work in industry, it’s easy to lose track of all the new buzzwords around data architecture, governance, metadata, and more. Fortunately, Prukalpa and Christine Garcia’s new deep dive provides a generous dose of clarity and perspective. They discuss the future of the modern data stack, evaluate recent trends, and extrapolate the practices and workflows data scientists are likely to adopt in the months to come.
- For AI research, 2023 might just be all about alignment. After a big year for large models and major advances in applied NLP methods, Tal Rosenwein and Guy Eyal reflect on what the next frontier might be in this thriving, well-funded ecosystem. Their overview covers emerging themes like reinforcement learning with AI feedback and alignment effectiveness, and touches on the growing importance of multimodal models. (If you ever feel overwhelmed by the relentless pace of AI research, don’t miss Thomas A Dorfer’s helpful tips on staying up-to-date.)
- An up-and-coming ML subfield continues to mature. Why not explore an exciting area of machine learning before the rest of the world catches on? Case in point: temporal graph learning. Shenyang(Andy) Huang and coauthors Emanuele Rossi, Michael Galkin, and Kellin Pelrine offer a comprehensive report on the state of the field, and look at the shape it will likely take this year.
After spending some time in the future tense, let’s not neglect the here and now—there are always new skills to learn and current developments to ponder. The year is off to a strong start here at TDS, with excellent articles on many fascinating topics. Here are some recent standouts.
- Yennie Jun shared an intriguing project about her own patterns of crying last year—it’s part personal reflection, part data analysis, and very much worth your time.
- The Statistics Bootcamp is back! We were thrilled to kick off the year with Adrienne Kline’s latest installment, which focuses on comparing two populations.
- If you’d like your R Markdown-produced reports to look more polished, Jenna Eagleson has a handy and concise tutorial to get you there in four steps.
- The arrival of ChatGPT has generated a lot of speculation about the future of search engines. Alberto Romero takes a close look at Google, Bing, and the stakes around this looming showdown.
- Make your in-house data tools user-friendly (and your colleagues happier and better informed) with a robust status page — Xiaoxu Gao explains how to go about creating one.
Thank you, as always, for supporting the work we publish. If you’d like to make the most direct impact, consider becoming a Medium member.
Until the next Variable,
TDS Editors
