Looking Back on Data Science in 2020
It’s hard to argue that 2020 was a strange year that led to paradigm shifts for almost everyone to a degree. Hygiene habits changed, work mostly moved virtual, and some industries had to reinvent how they look at their business models. For data science, some things may have gone unchanged, such as frameworks only getting more useful rather than seeing a change in dynamics. Other topics, such as ethical AI, have seen more major changes. What do a few practicing data scientists think were some of the biggest changes to come in the field of data science and AI in 2020?
Dr. Jon Krohn | Chief Data Scientist | untapt
DeepMind’s AlphaFold 2 algorithm being able to predict protein structure (e.g., the shape of the coronavirus protein Spike) to the same level of accuracy as X-ray crystallography, the manual, laborious, expensive, gold standard technique. This new algorithm leaves all previous computational approaches to protein-structure prediction in the dust by a staggering margin.
Violeta Mischeva | Data Scientist | ABN Amro Bank
Though GPT3 made quite a few headlines, the biggest development and most exciting moment for me was DeepMind’s leap in solving protein structures. Some have even labeled it one of the most exciting AI breakthroughs in years, and I agree with that.
Daniel Gutierrez | Data Science Consultant
2020 was the year of AI explainability/interpretability (XAI). Enterprises woke up to solving the “black box” problem with AI. Many start-ups received funding to develop technology to provide transparency for how algorithms make decisions. Here is a shortlist of some of the companies doing great work in this space: Beyond Limits, Darwin AI, explainX.ai, Fiddler, Kyndi, and Truera.
Jordan Bean | Senior Analyst, GRS North America Analytics | Liberty Mutual Insurance
I’m interested in the work that Tableau’s data science team has been doing behind the scenes with improvements to their AI-enabled tools like Explain Data and Ask Data. Their ability to blend a leading BI platform with robust AI and ML capabilities is a combination that should get data scientists excited to understand their data better and faster to ultimately improve model building and performance.
Francesca Lazzeri | Principal Cloud Advocate Manager, Cloud AI | Microsoft
The biggest development has been the ability to bring the latest research in responsible AI to the Cloud through open-source toolkits, such as FairLearn (www.aka.ms/FairlearnAI). This development has empowered data scientists and developers to understand machine learning models, protect people and their data, and control the end-to-end machine learning process. In order to make sure that machine learning solutions are fair and the value of their predictions easy to understand and explain, it was essential this year (2020) to build tools that developers and data scientists can use to assess their AI system’s fairness and mitigate any observed unfairness issues.
Kerstin Frailey | Data Science Manager, Data Quality | Numerator
Massive data disruptions. Most of us in the industry rely on data that is, in one way or another, directly driven by human behavior–and 2020 saw the greatest sudden changes in human behavior since the advent of our digital society. For all of us working with real-world data, the wild ride is only going to get wilder.
Dr. Kirk Borne | Principal Data Scientist and Executive Advisor | Booz Allen Hamilton
I guess a lot of folks will mention GPT-3. But, for me, I am excited to see the rapid growth in Observability as an important data strategy component in organizations. Having streams of data from sensors everywhere is awesome, and that’s called monitoring, which is something you do. Observability is why you do it — your data strategy — and that’s the most important conversation to have in data-intensive organizations.
Matteo Manica, PhD | Research Staff Member in Accelerated Discovery | IBM
Besides big splashes from large language models, the RoboRXN project was a great milestone and a poster child for AI acceleration of scientific discovery using Cloud and automation: https://rxn.res.ibm.com/roborxn. (I’m part of it so I might be biased.)
How to move on from data science 2020 trends
Between the new tools, expertise, and goals mentioned above, there’s a lot to learn about data science 2020 trends and beyond. 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.
Want to look ahead? Here are the above community members’ thoughts on the future of AI and their New Year’s resolutions.