What’s to come in 2021: 5 predictions for the future of data science and AI/ML
Since our founding in 2012, we’ve set out to create a movement that brings together data science practitioners, enterprises, and the open-source community. Data science has gone from a “nice-to-have” to a requisite for most businesses, and we’re proud to have witnessed its growth and expansion in recent years. But we also know there’s still much more to come.
That’s why, to start off the new calendar year, we asked a few of our experts here at Anaconda to look into their crystal balls and share what they expect to see in the data science space. Here are some of their predictions:
More tools focused on reducing and mitigating bias
The responsibility to create fair and explainable ML systems is greater than ever. In 2020, we saw many necessary conversations about bias in data and algorithms, and the harmful societal impacts it can cause. As we continue to discuss the implications and increase awareness of these issues, CTO Kevin Goldsmith says that 2021 will (and should!) see the continued development of tools that provide insight into the results of ML systems, reveal biases, and check drift in deployed models over time.
Understanding how to build an effective AI model marketplace
Building AI in-house and from scratch takes a lot of time and resources, and not all companies can afford that investment. That’s why Senior Director of Community Innovation Stan Seibert predicts that we will see the continued rise of “model marketplaces” that allow companies to integrate third-party AI for their needs. But even as we acknowledge the need for such marketplaces, there’s still a lot to figure out about how to do this most effectively. For instance, how do companies evaluate potential models for bias? How do they integrate the third-party models into their own services and infrastructure? How do businesses effectively monetize that ecosystem so that continued model development and maintenance are supported? All these questions will be important to address for effective model marketplaces moving forward.
The continued growth of MLOps and ML engineering
As AI/ML adoption accelerates, the role of ML engineers will be further refined, according to Data Scientist Albert DeFusco. Albert predicts that ML engineers will begin taking on more “operations” elements in their responsibilities similar to those we see in DevOps today. We’ll also see further development of tools for MLOps, whether to assist ML engineers in certain tasks or to automate parts of the pipeline entirely.
Improved synchronization across Python data visualization libraries
We’re finally starting to see Python data visualization libraries work together, and this trend will continue into the new year. Senior Technical Consulting Manager James Bednar notes that there has been variety and confusion in the past that have made it difficult for users to choose appropriate tools for visualization. In response, developers across organizations have been working to integrate Anaconda-developed capabilities, like Datashader’s server-side big data rendering and HoloViews’ linked brushing, into a wide variety of plotting libraries. This not only expands visualization capabilities for a wider user base, but also reduces the duplication of efforts, bringing a win-win situation for better and more seamless visualization.
Emphasis on open source’s value and the willingness to support the community
As enterprises have become more aware of the value of open-source software, Strategic Account Executive Matt Thornell says that we’ll see ties deepen between commercial users and open-source contributors. In fact, strengthening this relationship is one of the reasons we launched the Anaconda Dividend Program late last year. In 2021 and beyond, we’ll see more opportunities for commercial users to contribute back to the community and ensure the future sustainability of open-source software.
While these are only a sampling of the many advances that will come to fruition, we’re optimistic that practitioners, enterprises, and the community will move forward together in 2021. At its core, data science relies on a collaborative culture in which a rising tide lifts all boats, whether that includes advancement in open source tools or further research around deep learning algorithms. Here’s to a new year of growing with the dynamic field of data science!