COVID-19 response and recovery will be top of mind for nearly every firm and industry in 2021. Some industries may become stagnant or never recover. Others will view the shakeup as an unprecedented opportunity to understand and improve their data management, operationalize and update their model production process, and reassure customers that their AI can be trusted. Whether you are a bank improving fraud detection or anti-money laundering prevention, a healthcare provider shifting to telehealth, or a retailer or manufacturer working to make your supply chain more efficient, understanding how various industries are planning to use AI to bounce back from the pandemic can help your business thrive in 2021.  

Banking (Ryan Manville, AI Success Director at DataRobot)

As the world begins its recovery from the pandemic in 2021, dramatic swings will occur across the macro economy. A major theme will be the effects of fiscal stimulus and the reverberations that will be felt by households and larger companies. Banks and other financial institutions will be on the lookout for both substantial opportunities and significant threats, and the continual suppression of interest rates will be a major challenge as compressed spreads will weigh on profitability.  

Using outdated models will cause banks to rapidly lose profit, market share and, in some cases, reputation. Therefore, the ability to rapidly update models in areas such as underwriting, customer management, fraud, and collections—to name a few—will be paramount. 

Financial Markets (Peter Simon, Director, Financial Markets Data Science at DataRobot)

For the financial markets, 2020 turned out to be an unexpectedly good year, as the securities industry undoubtedly benefited from policy responses to the pandemic. We may well see some retrenchment from this in the year ahead, as volumes seem set to fade; therefore, an increased emphasis on cost control is likely. From an AI point of view, this means that we may see the following effects in 2021: 

• The focus on operational efficiency will increase — ”boring AI” will become ever more important on both buy- and sell-sides, with emphasis on less “sexy” use cases with measurable value: preventing STP failures, preventing AML false positives, and a whole raft of commoditized use cases in marketing.
• The need for agility will also increase, as conditions will continue to evolve quickly. This is where automated machine learning model development shines: using technologies such as DataRobot, the AI model development and governance lifecycle can be cut from months to days or weeks. 
• Data science practitioners will need to demonstrate their value to the business at an accelerating rate, or else they will struggle to justify their positions. The age of the AI “skunkworks” is over, with machine learning specialists being embedded in the lines of business and an emphasis on domain knowledge, not research.  
• Machine learning operations (MLOps) will become even more important as there will be an increasing focus on efficiently monitoring and robustly deploying AI solutions. “Data scientist code” just won’t cut it any more.
• And finally, increasing adoption of machine learning techniques by quants could give that sector some respite after three really terrible years.

Retail (Ari Kaplan, Director, AI Evangelism and Strategy at DataRobot) 

The retail industry is in crisis, but there are pockets of strength and opportunity, and the retail landscape will continue to see dramatic shifts in consumer behavior and the competitive landscape. Several uncertain factors will continue to challenge the industry in 2021: jobs, the economy, and the logistics of easing pandemic restrictions in respective regions. Retailers will be compelled to add AI into their business decisions—especially to understand the rapidly-changing, underlying data. MLOps will be a key focus of AI in retail to operationalize the model refresh process, detect drifts in consumer and economic data, and understand the importance of those changes.

Retail is the largest non-government employer in the U.S., comprising one in four jobs and 5.9% of the GDP — statistics that are similarly reflected around the globe. Retailers that adopt AI will be in a better position to address demand uncertainty and stabilization needs in the first half of 2021, obtain first-mover advantage in the long term, and later reimagine and reform their operations. AI will better connect to changing consumer needs, as stores move online and become more utilitarian, and retailers make better store location and repurposing decisions, optimize pricing, and change demand forecasting.

Healthcare (Jarred Bultema, Data Scientist at DataRobot) 

For many industries, 2020 has been a highly disruptive year, but perhaps none has experienced such a drastic range of changes as the healthcare sector. The financial impact of this is enormous—U.S. healthcare spending represents nearly 20% of U.S. GDP. 

The impact of the COVID-19 pandemic manifests itself differently for healthcare providers (i.e., hospitals, pharmacies, skilled care facilities, etc.) and healthcare payers (i.e., insurance providers). Healthcare providers have seen enormous disruption of their operating models, an inversion of revenue drivers and cost centers, new types of patients, and rampant illness among frontline providers. We have also seen the expansion of existing operational efficiency use cases, staffing forecasting, and staff attrition AI use cases. The onset of novel COVID-19-related use cases has also demonstrated tangible benefits to executive leadership.

In contrast, healthcare payers have seen shifting operational models and a reduction in claims volume. The direct impact of COVID-19 has resulted in a drastic reduction in costly elective procedures, new regulatory policies, and new pandemic-related funding streams that have altered the claims and payment landscape for health insurers. As the industry starts to view the risks to nursing homes and other palliative care facilities through a new lens, a shift and acceleration toward more home healthcare or telehealth will also likely be another indirect impact of COVID-19.

Healthcare leaders have told us that the tangible benefits and needs for AI within healthcare have never been greater. As norms are disrupted, we anticipate that whatever the new normal looks like in 2021 and beyond, AI will be a core capability as companies in the healthcare industry seek to implement both operational and healthcare-specific AI use cases.

Manufacturing (John Sturdivant, AI Success Director at DataRobot)

In 2021, manufacturers will still be responding to the shock of the new landscape created by the pandemic. Government and corporate policy responses to control the spread of the virus— coupled with dramatic shifts in underlying demand—will continue to force manufacturers to create agile, resilient, and cost-effective supply chains and manufacturing techniques. 

Machine learning will allow global manufacturers to both forecast demand for their products downstream and mitigate supply chain risks upstream. As the hard reset from early 2020 made older data less valuable for predictive purposes, AI and machine learning techniques will help build the best possible demand forecasts with this limited data pool. That is, machine learning models will be able to do more with less when predicting underlying demand for manufactured goods, whether they are children’s toys or steel ingots. 

By allowing global manufacturers to better anticipate and mitigate risks created by geopolitical, transportation, and economic trends, machine learning will also make sophisticated upstream supply chain management more accessible to the middle market without having to pay the seven-figure price tags typically associated with such solutions. This will be critical, given the volatility in global supply chains as countries and companies shut down or divert operations to deal with coronavirus.

Conclusion 

Industries are facing extraordinary opportunities and challenges in 2021. They need to navigate the post-COVID landscape, avoid AI bias, and deliver trusted AI to customers. Machine learning will allow global manufacturers to both forecast demand downstream and mitigate supply chain risks upstream, while the enormous disruption of healthcare operating models will be offset by the expansion of existing operational efficiency use cases, staffing forecasting, and staff attrition AI use cases. Whether it is underwriting, customer management, or fraud detection, models need to be functional, deliver real, AI-driven ROI, and do more with less data. The age of AI experimentation is over. The time to democratize and deploy robust AI solutions is now.

About the author

Peter Simon

Lead Data Scientist, DataRobot

Peter leads DataRobot’s financial markets data science practice and works closely with fintech, banking, and asset management clients on numerous high-ROI use cases for DataRobot’s industry-leading automated machine learning platform.  Prior to joining DataRobot, he gained twenty-five years’ experience in senior quantitative research, portfolio management, trading, risk management and data science roles at investment banks and asset managers including Morgan Stanley, Warburg Pincus, Goldman Sachs, Credit Suisse, Lansdowne Partners and Invesco, as well as spending several years as a partner at a start-up global equities hedge fund. Peter has an M.Sc. in Data Science from City, University of London, an MBA from Cranfield University School of Management, and a B.Sc. in Accounting and Financial Analysis from the University of Warwick.  His paper, “Hunting High and Low: Visualising Shifting Correlations in Financial Markets”, was published in the July 2018 issue of Computer Graphics Forum.

Meet Peter Simon

Ari Kaplan

AI Evangelist, DataRobot

Kaplan is a leading figure in data science, sports analytics, and business leadership. High profile roles include creating the Chicago Cubs analytics department, President of the investigation into the fate of Holocaust hero Raoul Wallenberg, and President Emeritus of the worldwide Oracle Users Group.

Meet Ari Kaplan

Jarred Bultema, PhD

Data Scientist, Time Series at DataRobot

Jarred is a multidisciplinary scientist with a diverse background. A trained Biochemist, Immunologist, and former University Professor prior to becoming a Data Scientist, Jarred also worked in the Biotech and Pharma industries. Jarred now works with clients across industries and governments to integrate sophisticated forecasting into their organizational DNA.

Meet Jarred Bultema, PhD

John Sturdivant

AI Success Director at DataRobot

He has led or advised CEOs in digital transformations across several industries and geographies. He lives in Dallas, TX with his wife and dog. Prior to joining DataRobot, he was Head of Digital and Transformation at TSS, LLC and a consultant at McKinsey & Co.

Meet John Sturdivant