2020 was a strange year and no one could have predicted what would happen. When we looked at trending topics around this time last year, we based it off of the usual trends and developments that we could have expected. With COVID and the new normal, data science and AI growth were also impacted. A larger emphasis has been placed on remote collaboration, and medical professionals have looked to AI to help with diagnosis. As we look to 2021, data science and AI professionals have varying opinions on what to expect for the year ahead, and here are some data science 2021 trends to keep an eye out for.
2020 was a big year for bringing the conversation around unwanted demographic biases into the mainstream of the ML community’s consciousness, but there’s a lot of hard work ahead to now resolve these widespread issues.
I am excited to follow all the new research in the area of XAI but I often witness a big gap between the theoretical research and what it needs for its application to be successful. Therefore, I hope to see more efforts in the practical side of XAI.
I’m hoping to see research that better documents the best-practices for causal inference techniques. While there is a wealth of knowledge published by accomplished researchers, there are many unanswered questions that are relevant in my work.
I’m hoping to see data scientists use their “data superpowers” in more ethical ways. No more deep learning for “deep fakes.” No more data scientists being fired and raided by police for standing up for ethics in data (e.g. see the plight of Rebekah Jones). I advise all my Intro to Data Science students that one day they may be asked to use their data skills for nefarious purposes and to be prepared to make an ethical decision.
I’d like to see more training and research into how to integrate decision science and data science. We’ve come so far in the capabilities of data science but the findings have been historically formed and delivered by individuals trained in coding and research rather than persuasion and influencing. People talk about the “unicorn data scientist” that is equal parts technical and business and I hope to see these individuals become more common in an organization.
With the release of Redshift ML and the continued maturity and adoption of BigQuery ML, I’m excited to see if anything happens in this space in open source, if and how data analytics continues to try to bleed into data science spaces, and if there are any other well-developed analytics technologies, that like data warehouses, can be used for data science workloads.
I hope to see more companies investing in Machine Learning Operations (MLOps) approaches, in order to improve the quality and consistency of machine learning solutions. MLOps could help data scientists by providing faster experimentation and development of models, and, most importantly, faster deployment of models into production!
True inclusion and accountability in AI, in data science, and in application. It’s been 4 years since Cathy O’Neil warned us all of the dangers already resulting from reckless, uniformed, and negligent use of machine learning. I’m tired of waiting for someday. I’m tired of seeing stories like what happened to Nijeer Parks. I want change, I want culpability, and I want consequences. And I want it now.
I am looking forward to publishing high-quality research papers in top tier conferences & journals and spreading the love of open-source in the community by helping & training more people. And I really would like to start traveling again ????
I think that a field that is very promising is the application of machine/deep learning to accelerate scientific computing simulations. Implementing methods to reach the proven theoretical convergence of neural network approximations for the solutions, can drastically change a wide variety of industries where R&D is heavily based on such simulations.
Best practices in operationalizing data science and machine learning have become just as important as software development and cybersecurity in 2020. I believe you will hear a lot more about “DevSecAIOps” initiatives in the enterprise in 2021.
In 2021, I am anticipating more edge computing developments, including on-chip machine learning (neural processors, anomaly and pattern detection processors), autonomous sensored devices (data-driven decisioning at the point of data collection, built into IoT and IIoT applications), and collaborative networks of such devices (collective intelligence at the edge, enabled by 5G network communications).
How to prepare for data science 2021 trends
Between the new tools, expertise, and goals mentioned above, there’s a lot to learn about data science 2021 trends. To get ahead, further education and training will help.
By signing up for an Ai+ Training Platform subscription, you’ll gain access to live and on-demand training sessions throughout the year, all of which focus on in-demand data science knowledge, core concepts and skills, and more.